Scanlon and Contractualism

Scanlon and Contractualism Editor MATT MATRAVERS University of York FRANK CASS LONDON • PORTLAND, OR First Publishe...

Author: Matt Matravers

35 downloads 991 Views 739KB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

Scanlon and Contractualism Editor

MATT MATRAVERS University of York

FRANK CASS LONDON • PORTLAND, OR

First Published in 2003 in Great Britain by FRANK CASS PUBLISHERS Crown House, 47 Chase Side, Southgate London, N14 5BP This edition published in the Taylor & Francis e-Library, 2005. “To purchase your own copy of this or any of Taylor & Francis or Routledge’s collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk.” and in the United States of America by FRANK CASS PUBLISHERS c/o ISBS, 920 NE 58th Avenue, Suite 300 Portland, Oregon, 97213–3786 Copyright © 2003 Frank Cass & Co. Ltd Website: www.frankcass.com British Library Cataloguing in Publication Data Scanlon and contractualism 1. Scanlon, Thomas. What we owe to each other 2. Contractarianism (Ethics) I. Matravers, Matt 171 ISBN 0-203-49867-4 Master e-book ISBN

ISBN 0-203-58446-5 (Adobe eReader Format) ISBN 0-7146-8456-2 (paper) ISBN 0-7146-5573-2 (cloth) Library of Congress Cataloging-in-Publication Data Scanlon and contractualism/edited by Matt Matravers. p. cm. Includes bibliographical references and index. ISBN 0-7146-5573-2 (alk. paper)— ISBN 0-7146-8456-2 (pbk.: alk. paper) 1. Contractarianism (Ethics) 2. Scanlon, Thomas. I. Matravers, Matt. BJ1500.C65S32 2003 171–dc22 2003015182 This group of studies first appeared as a special issue of Critical Review of International Social and Political Philosophy, ISSN 1369–8230, Vol.5, No.2 (Summer 2002) published by Frank Cass and Co. Ltd. All rights reserved. No part of this publication may be reproduced, stored or introduced into a retrieval system or transmitted, in any form, or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior written permission of the publisher.

Contents

Introduction: Scanlon’s Contractualism Matt Matravers

1

Scanlon and Reasons Sarah Marshall

11

The Magic in the Pronoun ‘My’ Susan Mendus

27

Contractualism, Spare Wheel, Aggregation Brad Hooker

44

Responsibility and Choice Matt Matravers

65

Promises and Perlocutions Michael Pratt

79

Contractualism and the Virtues Jonathan Wolff

103

Abstracts

114

Index

117

Books of Related Interest TRUSTING IN REASON: MARTIN HOLLIS AND THE PHILOSOPHY OF SOCIAL ACTION edited by Preston King THE PHILOSOPHY OF UTOPIA edited by Barbara Goodwin PLURALISM AND LIBERAL NEUTRALITY edited by Richard Bellamy and Martin Hollis THE IDEOLOGY OF ORDER by Preston King HUMAN RIGHTS AND GLOBAL DIVERSITY edited by Simon Caney and Peter Jones TOLERATION by Preston King

Notes on Contributors

Brad Hooker is Professor of Moral Philosophy at the University of Reading. His book Ideal Code, Real World: A Rule-consequentialist Theory of Morality appeared from Oxford University Press in 2000. He has also edited or coedited a number of books, including Truth in Ethics (Blackwell 1996), Morality, Rules, and Consequences (Edinburgh University Press 2000) and Moral Particularism (Oxford University Press, 2000). Sarah Marshall is studying for a PhD at the University of York. Her thesis is on the foundations of contemporary contract theory, and uses Scanlon’s contractualism, as presented in What We Owe to Each Other, as an exemplar of such theory. Matt Matravers teaches political philosophy, and is Director of the Morrell Studies in Toleration Project, at the University of York. He is the author of a number of papers in political and penal philosophy, of Justice and Punishment: The Rationale of Coercion (Oxford University Press 2000), and is the contributing editor of Punishment and Political Theory (Hart 1999). He is currently writing a book on responsibility to be published by Polity Press in 2004. Susan Mendus is Professor of Political Philosophy and a member of the Morrell Studies in Toleration Programme at the University of York. Her book, Impartiality in Moral and Political Philosophy was published by Oxford University Press in 2001. She is currently working on a book on Political Integrity to be published by Polity Press in 2004. Michael Pratt is an Assistant Professor in the Faculty of Law at Queen’s University in Kingston, Canada. He teaches and researches in the area of private law and private legal theory, and he has a particular interest in the law and philosophy of voluntary obligations. Jonathan Wolff is Professor of Philosophy at University College London. He is the author of Robert Nozick: Property, Justice and the Minimal State (1991); An Introduction to Political Philosophy (1996); and Why Read Marx Today? (2002). He is currently working on questions concerning the practical implementation of theories of egalitarian justice.

1 Introduction: Scanlon’s Contractualism MATT MATRAVERS

The publication of Scanlon’s What We Owe To Each Other in 1998 was a philosophical event. The book is the most sustained and comprehensive defence of contractualism since John Rawls’s A Theory of Justice. It represents the culmination of nearly 20 years of thinking by Scanlon, whose original views on the subject were given in his 1982 paper ‘Contractualism and Utilitarianism’. In that paper, Scanlon was particularly concerned to present contractualism as an alternative to utilitarianism and to emphasise the contractualist account of moral motivation as superior to the account of motivation offered in utilitarianism. What We Owe To Each Other, too, contains an account of moral motivation, but also much more. In it, Scanlon develops a fully fledged contractualist moral theory, rivalling others such as utilitarianism and deontology. This is an account of the property of moral wrongness itself as well as an account of the reasons we have to attend to that property. Scanlon’s ambition in What We Owe To Each Other, then, is grand, but there are two important constraints on the work. First, Scanlon is concerned only with ‘what we owe to each other’; that is with a specific, core, sphere of moral relations. Second, he does not think that resolving questions of moral wrongness and moral motivation requires taking on fundamental metaphysical questions or even traditional meta-ethical questions such as whether there are moral facts. ‘The question at issue’, he writes, is not a metaphysical one. In order to show that questions of right and wrong have correct answers, it is enough to show that we have good grounds for taking certain conclusions that actions are right or are wrong to be correct, understood as conclusions about morality, and that we therefore have good grounds for giving these conclusions that particular importance that we normally attach to moral judgments (Scanlon 1998:2–3). Throughout the book, Scanlon emphasises this metaphysical minimalism and his commitment to being faithful to the phenomenology of moral judgement. Indeed, accurately capturing how it feels to deliberate over moral issues is not merely an attractive aspect of his theory, it is often presented as the grounds on which his arguments should be preferred to the alternatives.

2 SCANLON AND CONTRACTUALISM

The desire to be true to the experience of arguing and deliberating over moral questions motivates the most significant shift in Scanlon’s position between his 1982 paper and What We Owe To Each Other. In the paper, Scanlon accounted for the motivational force of moral propositions in terms of a desire to act in ways that can be justified to others on grounds that they could not reasonably reject. As Scanlon now admits, such a desire-based account is unable to address the question of whether a person who lacks the desire has any reason to be moral, and it cannot explain how the lack of this desire can be accounted for as a moral fault (Scanlon 1998:7). In response to this, Scanlon has rethought his position, and in What We Owe To Each Other he argues that it is reason, not desire, that is basic in motivating persons to action. This is a significant shift, moving Scanlon from one ‘camp’ in accounts of practical reason (identified with Hume) to another (identified with Kant). Scanlon’s argument is given in the first chapter of the book and is examined in Sarah Marshall’s contribution to this volume. Reasons, Scanlon argues, are ‘primitive’ in the sense that they cannot and do not need to be reductively analysed into any more basic notion, such as desire. Their function is to support or undermine what Scanlon terms ‘judgement-sensitive’ attitudes, such as respect, fear and belief, which are responsive to our judgements of the evidence both for and against them. Desire then, according to Scanlon, should be understood as directing an agent’s attention towards reasons that she already has for (not) performing certain actions. It is not, in itself, the source of reasons, except in some few ‘trivial’ cases (Scanlon 1998:48). Motivation on this account stems from the recognition of a reason, rather the presence of a prior desire. Scanlon’s position is that when a consideration is judged to be a good reason to Ф, an agent will, ceteris paribus, form the intention to Ф, just as in the case of belief, the recognition of x as a good reason for believing that P, usually entails the agent coming to believe that P. Moreover, Scanlon argues, reason judgements are the ‘kinds of things’ (Scanlon 1998:70) that can be correct or incorrect—true or false—and this is supported by the phenomenology of such judgements. If, against a background of stable judgements, held by the vast majority of a community, and in accordance with Scanlon’s standards of correctness for reason judgements, we make the specific judgement that P, then we are justified in regarding that judgement for all intents and purposes as correct. So, the relationship among belief, desire and motivation is given by contractualism as follows: desire frequently directs our attention to reasons we have for holding certain attitudes and performing certain actions. Consideration of these reasons leads us to make judgements as to the strength of these reasons. A reason judgement can be true or false, and by its nature as a judgement about reasons can motivate agents to perform certain actions. In unpacking and examining this account, Marshall connects Scanlon’s switch to reason with two arguments both of which are important to his contractualist theory: that persons have reason to reach agreement and that they have reason to act morally, even where they desire not to do so. Marshall is broadly sympathetic

INTRODUCTION 3

to Scanlon’s project, but she questions whether he has done enough to displace desire as central in understanding motivation. In particular, she questions Scanlon’s analogy between belief and intention formation that is an important part of Scanlon’s case. Her conclusion is that Scanlon’s theory of reasons provides a sustained and detailed argument in favour of the idea that moral requirements are inescapable; moral reasons speak to agents even when they go unheard or unrecognised. As Marshall notes, Scanlon’s first chapter on reasons provides the foundations for his contractualism. The second and third chapters of What We Owe To Each Other continue this process, dealing with values and well-being. Scanlon is first concerned to demonstrate the inherent problems of teleological views of value, arguing that such perspectives assume that value is something that can, and should, be maximised or promoted, whereas this is not always the right response to value. This argument rests on Scanlon’s contention that the kinds of things that can be valuable are not just states of affairs, as is commonly supposed, but can include objects and people as well as more abstract things such as scientific achievement. He terms this account of value ‘the buck-passing account’, as it says that ‘being valuable is not a property that provides us with reasons. Rather, to call something valuable is to say that it has other properties that provide reasons for behaving in certain ways with regard to it’ (Scanlon 1998:96). Scanlon can therefore be seen as a pluralist about value in two ways: he recognises multiple sources of value, which are these other reason-providing properties, and a multiplicity of potentially appropriate responses to valuable things (cf. MacLeod 2001). This leads him to challenge another common view that well-being should be seen as a ‘master value’ from which all other values are ultimately derived. Contractualism is built on the judgement that human life is valuable, because humans have the capacity of reason-sensitivity, and the appropriate response to this value is to treat humans ‘in a way that recognizes the capacity of human beings, as rational creatures, to assess reasons and to govern their lives according to this assessment’ (Scanlon 1998:106). This means that we must offer reasons for our conduct to others—our actions must be justifiable to others—and this leads to Scanlon’s now-famous contractualist formula that ‘an act is wrong if its performance under the circumstances would be disallowed by any set of principles for the general regulation of behaviour that no one could reasonably reject as a basis for informed, unforced general agreement’ (Scanlon 1998:153). Through this principle, contractualism provides both an account of what morality is and a tie to the account of moral motivation. The basic moral category is wrongness. It is right that we should justify our actions to others, but we do this through learning which actions would be wrong, if performed, and we respect what we owe each other when we refrain from performance of these actions. It is important to see that this does not imply that those actions allowed by the contractualist principle are necessarily admirable. But, they are


nevertheless actions that can be justified to others, and this is as much as Scanlon thinks we should aim for. Scanlon argues, as part of his reason-based theory of moral motivation, that we have reason to want to live in unity with our fellows, and that to achieve this we must at least behave in ways that do not fail the contractualist test. However, the buck-passing account of value, his assertion that well-being cannot be a master value, and his appeal to the Millian idea of living in unity with our fellows, prompt the question of why, when people have many different values and interests in their lives, the moral should have priority for them in their practical deliberation. This is a question that is especially significant for Scanlon because of the phenomenology of moral deliberation. We experience the moral as having a certain force, as commanding obedience, and we experience it as often conflicting with our desires and self-interest. In the third essay in this volume, Susan Mendus discusses the question of priority: the question of why moral value should take priority over other values, such as the values of love and friendship. Mendus describes Scanlon’s account of priority and contrasts it with that given by Christine Korsgaard (Korsgaard 1996). Finding both unsatisfactory as complete accounts, she argues that combining elements of the two provides ‘a richer understanding of the relationship between morality and other values, and a better explanation of the priority of the moral’ (p.vii–viii). Mendus’s strategy is to identify two distinct questions that together make up the issue of priority. First, does impartialist morality drive out other values? Second, even if it does not, how can reasons of impartial morality have priority over reasons grounded in other values? These questions, Mendus argues, are asked by, and must be answered in, the first person. One theorist who is happy to allow impartialism to drive out other values is William Godwin. However, Godwin’s position is untenable and most contemporary impartialist accept the need to accommodate values such as love and friendship. This, Mendus argues, prompts the second question: why, in cases of conflict, moral values should (usually) trump other values. As Mendus puts it, ‘the defence of priority must…address the agent him/herself and it must show whether and why the reasons of morality, impartially understood, should take priority over reasons grounded in other values, such as the value of friendship’ (p.39). Characterising it as the ‘reductivist response’, Mendus describes Scanlon’s answer to this question. This is that the conflict between impartial and other reasons is illusory, because the reasons offered by other values are themselves grounded in moral reasons. The example Scanlon uses is ‘friendship’. He argues that real friendship requires that one friend recognise the other as a person in his/ her own right, worthy of respect, and to be treated as such. So, real friendship is not merely compatible with equal respect, but is grounded in a recognition of the other as worthy of such respect. Moreover, were this not the case in some instance of supposed friendship, then the relationship would be ‘too dependent

INTRODUCTION 5

on the contingent fact of affection’ (Scanlon 1998:164) to count as a genuine instance of the value of friendship. For both these reasons, then, Scanlon believes friendship properly understood is grounded in morality and so the conflict between moral and other values is illusory. Before assessing this answer, Mendus contrasts it with another, to which she gives the tag ‘the identity response’ and which she finds in Christine Korsgaard. Korsgaard defends the possible conflict between moral and other values, denying that the partial can be reduced to the impartial. Moreover, in cases of conflict she holds that the impartial need not trump: ‘there is no obvious reason why your relationship to humanity at large should always matter more to you than your relationship to some particular person’ (Korsgaard 1996:128). This is in part because, for Korsgaard, friendship may be an important part of one’s practical identity so to allow its demands to be trumped may be a matter of endangering one’s understanding of oneself and of who one is. As Mendus notes, these two responses do not seem to address the problem identified. The reductivist response denies that the conflict identified in the priority problem is a real one. The identity response grants that it is real, but denies that the moral must take priority. Nevertheless, Mendus thinks that there is value in both accounts. Deploying Ibsen’s A Doll’s House, she argues that the reductivist correctly captures the fact that friendship is moralised. Respect for the other as a person has an important place in any proper friendship. The identity account correctly captures the fact that in friendship and love there must be reciprocal commitment to interests that belong to both, not to each singly. It is important that the parties come to identify with projects and ideals that are theirs. Yet, if there are attractions in both the reductivist and identity responses, there are, Mendus argues, also flaws. The reductivist is not able to explain to the agent why what he feels as the demands of friendship should be trumped by impartial reasons even if he grants that his friendship may not be proper unless and until it is moralised. Moreover, the claim that all and only friendships grounded in morality are proper and that everything else is just ‘contingent affection’ is phenomenologically implausible. It is, as Mendus puts it, too high a price to pay to resolve the conflict identified in the priority problem. By contrast, the identity response speaks to the motivation of the agent, but in a way that also strains phenomenological credibility. For Korsgaard, the demands of friendship are important because they are connected to the agent’s practical identity. However, Mendus argues, this gives the wrong kind of account. Helping one’s friend is a matter of helping one’s friend, not of protecting one’s personal identity. In any case, the account is melodramatic. In most cases, failure to help one’s friends makes one feel bad, it does not threaten one’s self-understanding. In the final part of her essay, Mendus reflects on the priority problem in the light of the successes and failures of Scanlon’s and Korsgaard’s approaches. In doing so, she draws attention to an important feature of impartialist writing in general and of Scanlon’s account in particular. This is that morality—and living


in unity with our fellows—is itself related to our sociality and the reasons we have to live together with others in a certain way. Impartial morality, then, is not a constraint of which we would ideally be free (to pursue our unregulated passions), but is itself an expression of mutuality and sociality. If this is right, then the conflict that arises between friendship and morality is a case of noninstrumental attitudes that are essential to morality being undermined by the demands of morality fully articulated. In many ways, of course, this makes things more difficult. Given that a disposition to treat others non-instrumentally grounds morality, impartial theory can hardly ignore partial concerns when they conflict with morality’s demands. In response, Mendus argues, we should reverse the normal order of explanation: ‘rather than taking as basic that sense of mutual recognition which implies equal respect for others, impartialists should take as basic that sense of mutual recognition which implies ties of affection and concern for others’ (p.51). This does not resolve the priority problem—it does not tell us what to do—but it does explain the intractable nature of the conflict without collapsing the distinction between partial and impartial reasons and without itself collapsing into melodrama. Moreover, in recognising and explaining the intractability of the hard cases in which these conflicts appear it remains true to our moral experience. In explaining and defending the details of contractualism, Scanlon touches on a number of key issues: why justification of action should be by way of principles, the different perspectives of different agents, the requirements of generality and fairness, and whether the claims of the worst off should receive priority. However, one discussion here stands out: the contractualist treatment of the aggregation of moral claims. This issue, and the overall status of Scanlon’s contractualism, is taken up by Brad Hooker in the fourth essay in this volume. Hooker’s essay begins with a discussion of one of the changes identified above as having occurred in Scanlon’s thinking between 1982 and 1998: the narrowing of the scope of the account to what we owe one another. Hooker argues that this narrowing of the scope of the theory fits with Scanlon’s view of the rational force of contractualist injunctions. By this, he means to call attention to the way in which Scanlon uses contractualism to express the respect that is due to rational creatures. However, as Hooker emphasises, the force of contractualist moral injunctions is not grounded in actual consent or agreement. Rather, as noted above, the reason to do what morality requires is found in hypothetical agreement (agreement made in ideal conditions). Hooker turns his attention to the hypothetical nature of Scanlon’s contractualism and to the test, that Scanlon sets himself, of showing that the contractualist formulation of ‘wrongness’ accords with those of our moral convictions in which we have confidence. As Hooker notes, there is a ‘trap’ for anyone who tries to formulate a fundamental moral principle that meets this test. The trap is that this is achieved by presupposing those very moral convictions in developing the fundamental principle. This, of course, is a familiar objection to

INTRODUCTION 7

contract theories. The allegation is that the contractual argument is redundant; it spins like a spare wheel without bearing any weight or doing any work. Hooker first examines the spare wheel objection in Scanlon’s commitment to examining candidate moral principles ‘one at a time’ while holding many others fixed (that is, Scanlon rejects the idea that we should try to start from an empty, non-moral, point and then consider candidate sets of moral principles). Hooker finds Scanlon’s reasons for this strategy unconvincing, and as at times inviting the spare wheel objection. In Hooker’s view, Scanlon could avoid this objection if he were to adopt a ‘welfarist contractualism’, but as he rightly says, Scanlon rejects this move on grounds of both fairness and responsibility. This, in turn, leads Hooker to focus on Scanlon’s arguments concerning fairness (Scanlon’s views on responsibility are examined in Matt Matravers’s essay in this collection). These he finds strangely thin. In particular, Hooker thinks Scanlon’s identification of fairness with arbitrariness ‘makes fairness wholly parasitic on other moral concepts’. He contrasts this with a contractualist account of fairness which Hooker thinks occasionally suggested in What We Owe To Each Other, but never adequately developed. In the final part of his essay, Hooker turns to the question of aggregation. He argues that Scanlon is caught in a dilemma: either his contractualism cannot accommodate the idea that ‘numbers count’ and that the combined effects on many outweigh the effects on one, in which case it is ‘hugely counterintuitive in some cases’. Or, it can, but only at the risk of falling into the spare wheel trap. In short, what is missing is an account of the notion of moral relevance that is needed to accommodate this idea that is genuinely contractualist. Having established the foundations for, and the structure of, his contractualism, Scanlon turns in chs 6 and 7 to responsibility and promising. These chapters are examples of the contractualist formula in action and are examined in the fifth and sixth essays in this collection, by Matt Matravers and Michael Pratt. Matravers begins his discussion of Scanlon on responsibility by characterising contemporary liberal egalitarianism as caught in a dilemma. On the one hand, there is the demand to be fair with respect to reasonable pluralism, which is a form of fairness normally taken to require that modern liberal political philosophy eschew metaphysics. On the other, there is the demand to be fair with respect to choice and chance, which is a form of fairness normally taken to require that principles of justice counteract good and bad luck while leaving people with the outcomes of their choices. Following Cohen (1989), Matravers argues that resolving what counts as choice and what as chance seems to require invoking particularly controversial metaphysical arguments and thus that the liberal egalitarian cannot satisfy both demands of fairness simultaneously. Matravers then turns to Scanlon because it seems as if Scanlon might offer a route out for the liberal egalitarian. Scanlon distinguishes between attributive responsibility—judgements of whether a person is responsible for a given action in the sense that ‘it is appropriate to take [that action] as a basis of moral appraisal of the person’ (Scanlon 1998:148)—and substantive responsibility—


judgements about who should pick up the costs or benefits that result from some action. This distinction is an important one and it offers the liberal egalitarian hope of a way out of the dilemma described above because it separates the questions of whether we ‘really’ are responsible and of whether those who are responsible should necessarily pick up the bill. So, for Scanlon, someone who is lazy because of bad socialisation as a child is attributively responsible (it is legitimate to think of the person as lazy), but if he cannot hold down a job because of his personality then he is not to be held fully substantively responsible (it is not legitimate to deny him unemployment benefits). Moreover, these categories, and our judgements concerning them, flow from Scanlon’s contractualism and from his conception of persons as judgement sensitive, so they are, like Scanlon’s theory in general, metaphysically minimalist. Much of Matravers’s essay is concerned with examining Scanlon’s arguments for the compatibility of attributive and substantive responsibility with what Scanlon calls the Causal Thesis. His conclusion is that Scanlon does not provide a full enough account of substantive compatibilism. In other words, Scanlon does not explain how and on what basis we can continue to make people pick up the costs of their choices. However, Matravers argues that this is not a reason to dismiss the path out of the liberal dilemma suggested by Scanlon. Rather, it is a reason to engage more deeply with the notion of substantive compatibilism and with when we may judge a person to have had an adequate choice when deciding on a course of action. Pratt’s argument is similarly critical and similar in its conclusion that Scanlon’s contractualism may yet yield arguments not deployed by Scanlon himself. Pratt, too, begins by situating Scanlon’s argument in the wider field, in this case in the field of promising and promissory obligation. Pratt locates Scanlon’s theory in the ‘perlocutionary’ camp. Pratt’s main argument focuses on the Principle of Fidelity, which is proposed by Scanlon in order to explain how making a promise can obligate the agent who makes that promise. His claim is that the perlocutionary nature of this principle leaves Scanlon’s theory open to an objection of circularity that can be made against all such theories of promising. In short, the promisor is only obligated when the promisee is assured that the promised action will be performed, but the promisor is unable to give this assurance without first taking on the obligation to perform the requisite action. Having established this objection, Pratt considers two defences against it, one offered by Scanlon and one he offers on Scanlon’s behalf. Neither, Pratt argues, succeeds and thus he thinks the circularity objection ‘fatal’. Far from concluding there, however, Pratt goes on to argue for an alternative non-perlocutionary theory of promise that he believes both avoids the charge of circularity and is supported by Scanlon’s contractualist moral theory. Above, I said that Scanlon’s contractualism is an attempt at a moral theory to rival utilitarian and deontological theories. In choosing to cite utilitarianism and deontological theory I was following the norm yet, as Jonathan Wolff notes in the final essay, there is currently a resurgence of interest in virtue ethics. Wolff’s

INTRODUCTION 9

paper stands at a greater distance from Scanlon’s work than do the others in this collection, but it is important as a response to the contractualist theory that Scanlon offers. Wolff begins by examining the reasons for the decline in virtue theory that has only recently been reversed. He identifies a number of problems and focuses in particular on two: narcissism and priggishness. Virtue theory may appear narcissistic because it focuses on the moral agent rather than others (the objects of the moral agent’s moral behaviour). And it may seem priggish in that it requires the agent to be a particular type of person rather than requiring the performance of certain actions at certain times. As Wolff notes, the kind of person with which virtue theory has become associated is a person ‘possessing certain low church traits, such as chastity, abstinence and sufficient meekness to be a candidate for inheriting the earth’ (p.123). Wolff’s argument is that contractualism is well suited to provide a response to these problems and his essay is concerned to sketch a contractualist theory of the virtues. He begins his discussion by noting that the structure of contractualism is based on combining the idea that everyone is capable of having a conception of the good, and has good reason to pursue it, with the notion of being prepared to take the interests of others into account when pursuing that conception (Rawls’s ‘rational’ and ‘reasonable’ moral powers respectively). If a life containing the virtues is to be chosen in the contract situation, then, it must be appealing as everyone is entitled to attempt to live a life that she finds appealing. This, of course, means that we must shift our attention to whether, and if so in what way, a life containing the virtues is an appealing one. Wolff argues that it is and his arguments share something with those of Susan Mendus for both argue that the appeal of a life containing the virtues is not only instrumental. Such a life may well be instrumentally valuable—social norms reward virtuous behaviour and the virtuous are more likely to find themselves included in social co-operation than are the vicious—but it may also be more than that. Much of Wolff’s paper is an extended reflection on, and defence of, the claim that a life containing the virtues may have appeal in its own right, an intrinsic non-instrumental value. Wolff’s essay, then, is not an attempt at a full scale contractualist theory of the virtues. Rather, he sketches the outline of what such an account would look like. In doing this, he achieves two things: first, he brings together two of the great schools of moral theorising (both of which are currently enjoying much attention) and thus locates Scanlon’s contractualism in the broader field of moral theory. Second, he shows how a contractualist theory of the virtues can respond to the objections of narcissism and priggishness. Both objections are shown to be one-sided. A life lived in attention to the virtues is not narcissistic because, although it is chosen because it is appealing to the agent, it is also chosen because of all appealing lives it is the one that gives the greatest scope for taking the interests of others into account. Similarly, it is not priggish because although it does require taking others into account, a virtuous life is appealing in its own right (it need not entail sacrifice and boredom).


Taken together, these essays not only show the importance of Scanlon’s work and of the contractualist theory more generally, but they make an important and original contribution to the ongoing debates of which they are a part. ACKNOWLEDGEMENTS In writing this introduction, I have borrowed from notes and memos passed between and among Sarah Marshall, Susan Mendus, and me whilst we were thinking about the nature of the collection. I have also borrowed from an earlier paper of mine (Matravers 1999–2000). I am grateful to the editors of Imprints for allowing me to use this material. The inspiration for this project came from a Morrell Conference at the University of York. As ever, I am grateful to the trustees of the C. and J.B.Morrell Trust for their support. REFERENCES Cohen, G.A. 1989. ‘On the currency of egalitarian justice’. Ethics, 99/4, 906–44. Korsgaard, C. 1996. Sources of Normativity. Cambridge: Cambridge University Press. MacLeod, C. 2001. ‘Making moral judgements and giving reasons’. Canadian Journal of Philosophy, 31/2, 263–90. Matravers, M. 1999–2000. ‘What’s in it for me?’, Imprints, 4/2, 163–72. Scanlon, T. 1982. ‘Contractualism and utilitarianism’. Sen and Williams 1982:103–28. 1998. What We Owe To Each Other. Cambridge, MA.: Harvard University Press. Sen, A. and B.Williams, eds. 1982. Utilitarianism and Beyond. Cambridge: Cambridge University Press.

2 Scanlon and Reasons SARAH MARSHALL

In the first chapter of What We Owe to Each Other (Scanlon 1998), Scanlon presents a theory of reasons, which forms the foundation for the contractualist theory that follows. This differs from his earlier contractualism (Scanlon 1982), which was based on desire. It is therefore important to understand the details of Scanlon’s views on reasons, because this will shed light not only on the virtues of a reason-based contractualism, but also help to uncover Scanlon’s perspective on the deficiencies of a desire-based account. Since reasons are foundational to Scanlon’s contractualism, it is also important to identify any problems with his account of reasons, as any weaknesses are likely to afflict the theory as a whole. I propose to split Scanlon’s argument concerning reasons into three parts. The first part discusses the nature of reasons, and the relation between reasons and rationality. The second part examines Scanlon’s claim that desires can neither motivate action, nor (usually) function as the source of reasons, and it will be argued that this amounts to an argument against a subjective component in reason judgements. The third part returns to the nature of reasons, proposing that reason judgements are, in Scanlon’s view, objective. I further argue that the overall aim of Scanlon’s account of reasons is to establish the objectivity of reason judgements in order to make the claim that individuals can and do have reasons to hold certain attitudes and perform certain actions, whether or not they recognise these reasons. This allows Scanlon to go on to make two claims: that individuals have reason to reach agreement and that they have reason to act morally, irrespective of their desires.1 Reasons and Rationality Scanlon begins his account of reasons by stating that he will take the idea of a reason as ‘primitive’. By ‘primitive’ he means that the concept of a reason neither can nor needs to be satisfactorily explained in terms of any more basic notion. In support of this claim he argues that any explanation of our ordinary understanding of reasons will lead to the idea of a consideration that counts in favour of something. But if we then ask ‘Counts in favour of how?’ the only apparent answer is that the consideration provides a reason. Thus, the idea of a


reason as we ordinarily understand it appears to be a basic element of practical reasoning. As is implied by this initial definition of a reason, Scanlon is only concerned with reasons in what he terms the ‘standard normative sense’ (Scanlon 1998:19), which are considerations that have normative force. Reasons that explain causal, empirical phenomena are not included in his account. So, for example, reasons why the volcano erupted are not included, because they concern causal physical occurrences and laws; but reasons why the volcanologist thinks the volcano erupted are included, because they comprise considerations in favour of a particular conclusion on his part. Reasons in this standard normative sense, then, can be seen to provide justification for beliefs and attitudes. Scanlon terms this class of attitudes, which includes belief, ‘judgement-sensitive attitudes’ (henceforth JSAs), and defines them as attitudes that an ideally rational person would come to have whenever that person judged there to be sufficient reasons for them and that would, in an ideally rational person, ‘extinguish’ when that person judged them not to be supported by reasons of the appropriate kind. (Scanlon 1998:20) They can also be more loosely defined as evaluative attitudes, such as belief, fear and respect. However, not every attitude with evaluative content is a JSA. For example, no type of phobia can be a JSA, because phobia persists in the face of considered judgements that the attitude is irrational and is thus not judgement sensitive. Scanlon notes that actions seem to be an exception to this claim about reasons, because they are not attitudes, and yet reasons can usually be given both for and against them. But, he argues that when actions are intentional, they constitute expressions of JSAs. For example, when David puts on a helmet before riding his bicycle, this may be an expression of his fear of sustaining serious injury to his head in the event of an accident, and of his belief that a helmet will protect him against such injury. So, Scanlon’s account of reasons is limited to reasons in the standard normative sense, which can be given for JSAs both directly, and as expressed through intentional action. This account of reasons is also reliant upon Scanlon’s views on rationality, which are explained in terms of reasons and JSAs. He asserts that ‘a rational creature is, first of all, a reasoning creature—one that has the capacity to recognise, assess, and be moved by reasons, and hence to have judgementsensitive attitudes’ (Scanlon 1998:23). Rational creatures will usually come/ cease to hold JSAs through careful deliberation about reasons, but will also often form certain JSAs unreflectively. In this latter case, formation of JSAs is generally constrained by judgements made previously, regarding the adequacy of reasons. This means that rational creatures will judge certain types of evidence to be more reliable than others, on the basis of experience. Thus, such creatures are, for example, more likely to believe the opinion of an expert than that of someone who knows little about the relevant issue.

SCANLON AND REASONS 13

Irrationality, then, ‘in the clearest sense occurs when a person’s attitudes fail to conform to his or her own judgements’ (Scanlon 1998:25), which means that an individual who judged there to be sufficient reason for not doing A, and did A anyway, would be irrational. This is uncontroversial, but the scope of the term ‘irrational’ is frequently taken to be wider than this. Its meaning is often extended to descriptions of individuals who continue to hold an attitude in the face of what seems to others to be an obviously relevant reason against it. Scanlon argues against this wider use of ‘irrational’ on the basis that it is not necessarily irrational to fail to recognise a reason. Such failure may, for example, be due to limited time for deliberation, or flawed prior reasoning, none of which obviously indicates irrationality. In this type of case, Scanlon suggests that the individual’s attitude may be open to rational criticism, while not being irrational. Thus, a grey area of rational criticism is left between rationality and irrationality; they are not simply flip sides of the same coin. Two things a theory of rationality might be expected to do is to tell us in what way we must think if we are to avoid irrationality, and to tell us what is the most rational thing to do. Scanlon has already given an indication of the former with the assertion that our attitudes must be consistent with our considered judgements, and this will be elaborated upon below. Accounts of the latter are often termed ‘conceptions of rationality’, but Scanlon argues that these are more accurately seen as ‘substantive conceptions of the reasons that we have’ (Scanlon 1998:30). A full account of the most rational thing to do would have to involve possession of perfect information regarding the situation and possible consequences, recognition of all relevant reasons, and perfect deliberation regarding what these reasons support. Scanlon implies that this kind of abstract ideal theory is not a great deal of use in practice. As he says, ‘given the great variety of reasons for action, it seems to me very unlikely that there could be such a thing as a theory of reasons in this sense—that is to say, a systematic, substantive account of what we have most reason to do’ (Scanlon 1998:32). Furthermore, we can never know if we have full information, or if we recognise all of the relevant reasons, and therefore whether or not our reasoning is free from error. On this basis it seems that any attempt to discover the most rational thing to do or think in a particular situation must of necessity be an approximation to the ideal. What is of greater use in such situations is the notion of reasonableness. Scanlon states ‘judgements about what it is or is not reasonable to do or think are relative to a specified body of information and a specified range of reasons, both of which may be less than complete’ (Scanlon 1998:32). Reasonableness therefore uses the idea of the most rational thing to do or think in a more realistic sense. For example, it is reasonable for Catherine to judge that the Japanese tourist wants directions to the station if he is pointing at the station on a map and looking confused about his present location, even though he is speaking Japanese and Catherine does not understand Japanese. Conversely, it is also reasonable for Catherine, as a layman, to believe the results of a flawed scientific study if she is


unaware that it is flawed and information concerning its faults is not widely available outside the scientific community. This second example highlights another important point about reasonableness: while the reasonableness of an attitude is often assessed relative only to the agent’s beliefs and the reasons she sees as relevant at the time, it is also frequently assessed on broader grounds, so long as the relevant information and reasons are (fairly easily) available to the agent.2 This would usually be when agents are reasoning from the same body of information and reasons, such that an appeal to be reasonable could be a criticism of the way a particular agent is reasoning, but could also be a request to take a certain fact into account, which is currently being ignored. This dual sense of reasonableness plays an important role in Scanlon’s later arguments concerning hypothetical agreement. The Nature and Role of Desire Desire has often been thought of as an extremely important element in practical reasoning. Some theorists have argued that desire (in some form) is the source of all of our reasons and/or the motivating component in our reasons. Other theorists have argued that desire has a smaller role to play, and Scanlon takes this latter course.3 He begins his analysis of desire by considering motivation, arguing that reasons alone can be motivationally efficacious and that there is therefore no need to postulate an additional motivating factor such as desire. This argument starts from an analogy between belief and intention formation: A rational person who judges there to be sufficient grounds for believing that P normally has that belief, and this judgement is normally sufficient explanation for so believing. There is no need to appeal to some further source of motivation such as ‘wanting to believe’. Similarly, a rational person who judges there to be compelling reason to do A normally forms the intention to do A, and this judgement is sufficient explanation of that intention and of the agent’s acting on it (since this action is part of what such an intention involves). There is no need to invoke an additional form of motivation beyond the judgement and the reasons it recognises, some further force to, as it were, get the limbs in motion. (Scanlon 1998:33–34) Scanlon claims this follows from his account of rationality. In the case of belief, this is indeed precisely what was claimed in the account of rationality described above. However, in the case of action things are less straightforward. When an agent recognises a compelling reason to do A, this does seem to be a sufficient explanation of his subsequent intention to do A, but what if he does not do A? For example, Ashley may recognise the fact that his credit card bill is due for payment at a particular time as a compelling reason to pay it by the due date, but this does not necessarily mean he will pay it by the due date. He may have enough money and ample opportunity to pay the bill. He has no good reason not


to pay it, but even this, in conjunction with his recognition of a compelling reason to pay the bill, does not guarantee that he will pay it. When an agent acts on an intention such as this, his recognition of a compelling reason may seem to explain both the intention and the action, but when he does not so act further explanation is required, which casts doubt on the sufficiency of the explanation for the former case. This problem suggests that the analogy between belief and intention formation is suspect, as the distinction between recognising a consideration as a reason for action and being motivated by it to the extent that the action is performed does not seem to apply in the case of belief. It appears impossible to think that X is a good reason for believing that P and not believe P in the absence of stronger contradictory reasons, but it does seem possible to think that X is a good reason to do A and not do A, even without any good reason not to do A. In order to maintain the analogy between belief and intention formation, Scanlon argues that agents can sometimes have ‘bad’ reasons for belief. These are reasons for belief that are acknowledged by the agent in question to be overridden by conflicting reasons, but still, contrary to the presumption above, function as compelling reasons for that agent. An agent forming beliefs on the basis of such ‘bad’ reasons would be demonstrating akrasia of belief. To support this contention, Scanlon gives the example of Jones, the false friend, whom he knows to be an ‘artful deceiver’, but whose pretensions to being a loyal friend are nevertheless to some extent convincing, leading Scanlon to believe that Jones can in fact be trusted (Scanlon 1998:35). One problem with this alleged akrasia of belief may be thought to lie in the fact that it is impossible to believe something you simultaneously know to be false. In the above example, Scanlon is supposed to know that Jones is an artful deceiver, and yet believe that he may be reliable after all. But Scanlon is not claiming that two conflicting beliefs are held simultaneously; rather, the claim is that one belief, based on good reasons, is (temporarily) ditched in favour of one based on bad reasons. The agent holds a particular belief against his better judgement. Is this plausible? One kind of case in which it might be is that of the breakdown of a relationship. In such cases there is often one agent who finds it difficult to accept that the relationship is over, and despite claiming to ‘know it’s over’, actually believes that there is a chance of reconciliation, perhaps because he remains friends with the former partner, and remembers their previous expressions of undying love. This appears to fit the model of akrasia of belief, as the agent holds a belief against his better judgement. But, this kind of case introduces a further element, which is the strong possibility that the agent is deliberately deceiving himself—he wants to believe that the relationship may not be over, even though he knows it is. In holding the ‘wrong’ belief, he is in fact indulging in self-deception.4 Thus, in this kind of case, there is a need to appeal to a further source of motivation, that of ‘wanting to believe’. If the aim here were simply to establish whether akrasia of belief is possible, then the example of relationship breakdown would help considerably, as it


captures not only the idea that the agent is holding a particular belief against his better judgement, but also the specific notion of weakness of will. The agent’s rational judgement is overcome by his desire for the world to be different, by his desire for the bad reasons to be good ones. However, an analogy between belief and intention formation on this basis is problematic for Scanlon. The whole point of the analogy is to undermine the view that desire motivates action, by saying that reasons operate as the sole motivating factor in both belief and intention formation. But, the self-deception interpretation of akrasia of belief suggests that this analogy can only be sustained through the attribution of an additional motivating factor, such as desire, to both cases. This implies that Scanlon must either claim that cases of self-deception are not true examples of akrasia of belief, or drop the analogy between belief and intention formation. It is of course true that the Jones example, as stated, is not a case of selfdeception, and therefore it is still possible that akrasia of belief could occur without self-deception, sustaining the analogy required by Scanlon’s argument. But, it is also true that the Jones example, as told, is very thin. It runs contrary to the more generally accepted interpretation of the phenomenology of belief formation—that there is no akrasia of belief—without providing any feature in virtue of which it can be reconciled with the more usual interpretation. It seems more plausible to argue either that Scanlon temporarily regarded Jones’s genuine appearance as a good reason for doubting his judgement of Jones as an artful deceiver, or that Scanlon wanted to believe for some reason that Jones’s appearance reflected his real character. Perhaps the Jones example can be made plausible, but much more information is needed if it is to carry the necessary weight in the argument. That said, the provision of such information would be likely to damage rather than strengthen the example, by implying that the situation in fact conformed to one of the more plausible types of case given above. Nevertheless, these claims rest on an appeal to a different interpretation of the phenomenology of belief formation from the one appealed to by Scanlon, and as such can only cast doubt on Scanlon’s own interpretation and the validity of the Jones example. They cannot show Scanlon’s account to be false. My argument is that the analogy between belief formation and intention formation rests on shaky ground. There are good reasons to think that this analogy does not hold, at least in the way it is thought to by Scanlon. This means that Scanlon does not have firm grounds for using the absence of an additional motivating factor, such as wanting to believe, in the case of belief to show that it is unlikely there is an additional motivating factor, such as desire, in the case of intention and action. If the analogy does not hold, then although there is usually no additional factor in the case of belief, this has no bearing on whether or not there is usually an additional motivating factor in the case of intention and action. The two could simply be seen as different in this respect. To summarise: my claim is not that there is always an additional motivating factor for the formation of intention and action. Sometimes the recognition of a compelling reason does seem to be enough to motivate action. My claim is simply that, as regards


intention and action, there are many cases in which an additional factor does seem to be required, whereas with belief the necessary presence of any additional motivating factor appears to be very much the exception rather than the norm. So, given the weakness of the analogy, Scanlon must provide further argument to show that desire plays no role in motivation. Interestingly, Scanlon does give further argument in the form of a detailed analysis of the concept of desire. He begins by stating an apparent truism about desire:5 1. Desire is supposed to be motivationally efficacious. 2. Desire is supposed to be normatively significant, in the sense that when an agent has a reason to do A, this is generally true because doing A would promote the fulfilment of one of his desires. This apparent truism holds when desire is conceived of very broadly, to include any ‘pro-attitude’ of the agent. Even cases where it is natural to say that the agent had no desire to do A and did A anyway can be explained by a pro-attitude interpretation of desire. For example, Kate does not want to give Dan some bad news, but she feels that she should, and so does. Kate’s action here can be explained by her feeling some duty or loyalty towards Dan, so she can actually be seen to have a stronger pro-attitude towards telling Dan the news herself than against this action. It is also plausible to claim here that such pro-attitudes are the only things capable of motivating intentional action, as it seems that any intentional action could be susceptible to a pro-attitude interpretation. Scanlon objects to this pro-attitude conception of desire on the grounds that many such attitudes are not original sources of motivation. Rather, they seem to be the motivational consequences of the agent’s recognition of a consideration as a reason (he terms these attitudes ‘motivated desires’, Scanlon 1998:37, following Nagel 1970). In the example above, Kate’s strongest pro-attitude is a sense of duty or loyalty, and surely these are just judgement-sensitive attitudes. They have evaluative content, and would presumably, in rational agents, be adopted on the basis of compelling reasons in their favour and in the absence of compelling reasons against them. So, Kate’s pro-attitude is actually the outcome of deliberation about reasons, and it is therefore more appropriate to say that it is these reasons that supplied the motivation rather than the pro-attitude itself. This means, according to Scanlon, that any theory assigning desires a special role in motivation would have to be based on a narrower conception of desire, one that included only ‘unmotivated desires’. Scanlon considers thirst as an example of an unmotivated desire, breaking it down into three components: the sensation of having a dry throat, the belief that some action (in this case having a drink) would lead to a pleasant state in the future, and taking future good to be a reason for acting (Scanlon 1998:38). He argues that the desire part of thirst seems to consist in an urge to drink, but the motivational work is done by the agent taking the future pleasure to be obtained


by drinking as a reason for action. This is because the urge alone cannot rationalise action. While an integral part of desire may be some form of urge, another such integral factor is evaluative. A desire for something seems necessarily to involve seeing something desirable about it. After all, if we did not see anything desirable about the things we desire, it would seem strange to say that we do in fact desire them. This suggests that it is not even unmotivated desires that motivate us; rather it is our reasons for finding their objects desirable that supply the motivational spur to action. Desire itself seems to be motivationally inert. So, where does this leave desire as a concept? How exactly does Scanlon think of desire? He claims that the conception of desire that fits best with our ordinary notion of desire is that of ‘desire in the directed-attention sense’. An agent has a ‘DA desire’, as I shall term it, ‘if the thought of P keeps occurring to him or her in a favourable light…[his] attention is directed insistently toward considerations that present themselves as counting in favour of P’ (Scanlon 1998:39). This fits with our ordinary conception of desire in including the appetitive and the evaluative elements identified above as essential. DA desire, as with ‘ordinary’ desire, can also conflict with our considered judgement regarding what we should do, and in persisting to present a consideration as a reason even after the judgement has been made that it is not a reason, can also be irrational.6 DA desire effectively captures the kind of desire felt for objects that we do not really need, or for things we would not do if we were completely rational. It is true that considerations like ‘I don’t own a pair of pink sandals’ can continue to appear to be reasons for buying a pair of pink sandals even when the decision has already been made that such sandals are not needed and would be a waste of money. Conversely, DA desire can also affect our deliberation and subsequent actions by bringing valid reasons to our attention that we do decide to act on. However, DA desire does not accurately capture other types of desire. Longterm desires, such as career ambitions or ‘things to be done before I die’, do not usually keep coming to mind in this way, but they do involve an urge to do something and the perception of something desirable about it, and therefore seem to be desires in the ordinary sense. Other desires can be latent rather than ‘on one’s mind’. For example, John may ask Greg if he wants to go to the pub, and Greg may say yes because he wants a beer. But, the thought of having a beer may not have occurred to him until John suggested going to the pub. In this case, it would be true to say that Greg went to the pub because he wanted, or desired, a beer, but it does not seem appropriate to say that he had a DA desire for a beer. These two examples show that the notion of DA desire does not do justice to our ordinary conception of desire. It gives too narrow an account of the nature of desire, and cannot therefore be considered sufficient as an interpretation of the concept of desire. This means that we should be cautious when dealing with arguments that take DA desire to represent the concept of desire as it is ordinarily understood.


This point does not render the above analysis of desire worthless. What has been established is that the structure of desire, in including both appetitive and evaluative elements, cannot motivate in the way that some theorists have previously assumed. The argument is that it is our reasons for finding the objects of our desires desirable that motivate us to act. As was shown by the drink example, Louise desires a drink because it will stop her being thirsty, which reason shows how the drink as the object of Louise’s desire may be thought desirable. This gives her reason to drink, not because she desires a drink (although, of course, she does), but because it will slake her thirst. Desire seems to add neither to the perception of a reason nor to the subsequent formation of an intention to act on that reason (and subsequent action). Desires do not directly motivate action; on the basis of the above analysis, reasons do. So how can this conclusion be reconciled with the conclusion of the argument concerning the analogy between belief and intention formation (that we often need a reason and/or an additional factor to motivate action)? What is clear is that desire does not motivate in the way it has often been thought to, and DA desire does not seem capable of directly motivating action. But, this does not show that desire plays no role in motivation. It is possible that Scanlon’s conception of desire as DA desire is too limited to capture the precise way in which desire does motivate, making it appear that it is only reasons that have motivating force. Scanlon continues his analysis of the role of desire in practical reasoning by examining the claim that reasons for action can, or must, derive justificatory force from desires. He says that the standard case of having a reason for action is generally held to be described by means-end reasoning (part two of the truism above), and notes that there has been much debate concerning whether or not all reasons for action are based on desire in this way. Some kinds of reasons are clearly dependent on desires in a way that is unproblematic. For example, reasons concerning which flavour of ice cream to choose, or where to go on holiday, depend on the subjective reactions of the agent in the sense that the subjective reaction constitutes the reason. However, other types of reasons, specifically moral and prudential reasons, do not seem to be dependent on desire. To say that they are so dependent implies that ‘the requirements of morality and prudence [are] “escapable” in a way that they clearly are not’ (Scanlon 1998:41– 2). In order to defend the claim that moral and prudential reasons are not dependent on desire, Scanlon makes the stronger claim that desires almost never provide reasons for action in the way suggested by the means-end model of reasoning above. This claim is supported by consideration of whether DA desire could be a source of reasons. On the face of it, DA desire does not seem to be a likely candidate for the role of a source of reasons because this type of desire has evaluative content. The reasons that a DA desire highlights are those that are prior to the desire itself—those that support the idea that the object of the desire is indeed desirable.


However, Scanlon points out that DA desire does seem to add something to general enjoyment. For example, an agent may generally enjoy raspberry muffins, but that is not enough to give rise to a reason now to procure such a muffin. Whereas the specific DA desire for a raspberry muffin does supply such a reason. But if DA desires are not original sources of reasons, how could this be? The crucial factor here, for Scanlon, is that of ‘taking a consideration to be a reason’. General enjoyment of raspberry muffins does not always provide a reason because it is by no means constantly taken to be a reason by the agent. This may change if the agent walks past a muffin shop while feeling a little hungry, because in this situation, her general enjoyment of raspberry muffins may come to mind in the form of a DA desire, and she may take this to be reason, at that time, to buy a raspberry muffin. Her desire and her reason to fulfil the desire occur simultaneously because the same consideration, that of enjoyment of raspberry muffins, operates both as the reason why the object of the desire is desirable, and as the reason to fulfil the desire. This is why it appears almost impossible to have a DA desire without recognising a reason to fulfil it, even if on reflection the reason is judged to be a bad one. This shows that there are very few cases in which a desire may be thought to be the source of a reason, which seems to be because desire, as in DA desire, has evaluative content. DA desire cannot be a source of reasons because it is not itself independent of our practical reasoning. Subjective reactions are independent of practical reasoning, and this is why many reasons have subjective conditions. For example, Claire has reason to buy a raspberry muffin because she prefers raspberry muffins to chocolate ones. But, these subjective reactions are not themselves desires, at least on Scanlon’s DA desire model, because they do not have evaluative content; and they are not usually sources of reasons because, without evaluative content, they cannot alone rationalise action.7 They are usually preconditions for reasons being reasons for particular agents, not reasons in themselves. For example, that a career in medicine is worthwhile is a reason to pursue that career, but it is only a reason for Naomi if she feels attracted to a career in medicine. Thus, it is clear that desires cannot be sources of reasons due to their evaluative content (reasons are prior to desires), and subjective reactions are argued by Scanlon only rarely to operate as sources of reasons. If the latter claim is right, then the standard means-end model that says reasons for action are justified by desire is misleading at best. Scanlon’s claim that desires almost never provide reasons for action will have been shown to be defensible. If Scanlon’s claims about desire were valid, then he would have established that desire does not play either of the two roles it is commonly thought to play. This conclusion would remove a large degree of subjectivity from practical reasoning. It would tell us that rational agents will recognise reasons, and that these reasons justify their attitudes and actions, while also motivating those actions; agents would not usually be thought to act simply because they desired certain states of affairs. However, final judgement on this conclusion must be suspended because of the doubts canvassed above over Scanlon’s claims regarding the role of desire


in motivation. Scanlon’s claim that desire cannot be the source of reasons appears more robust. But, as it stands, it is still possible that desire could supply a significant subjective component in practical reasoning. The Objective Nature of Practical Reasoning Scanlon moves on to offer some positive arguments for his view that reason judgements are objective in nature. To this end, he provides a more detailed definition of reasons. He states that reasons are ‘the sort of things, picked out by “that” clauses, that are the contents of beliefs’, for example, ‘that the hat is day-glo pink is a reason not to buy it’ (Scanlon 1998:56). These ‘things’ are propositions, but Scanlon does not mean that reasons are simply propositions; rather he means that reasons are propositions that additionally contain an element of normative force. In this way, they resist ‘identification with any proposition about the natural world’ (Scanlon 1998:57). Scanlon argues that in order for a proposition to have normative force for a particular agent, the agent must believe it, and must take it as a reason for the relevant attitude. This could imply that propositions gain normative force through the way in which they are perceived by agents. But what grounds the ‘taking it as a reason’ part? On what basis do agents make the judgement that X does indeed count in favour of holding the attitude, or performing the action, A? Before answering this question, Scanlon attempts to answer the prior question of how we should interpret the judgement that a particular consideration counts as a reason for something. He argues that there are two possible ways: as a belief that the relation of ‘counting in favour of’ holds between X and doing A; or as a special JSA of taking X to count in favour of doing A. Superficially, the latter is more plausible because of three objections to which the former is liable: first, the belief account would have difficulty in explaining the presence of normative commitment to a particular conclusion, as this seems to be missing from a simple judgement of fact. Second, the belief in question would most likely be a belief in a kind of non-natural fact, which could be metaphysically problematic. Third, it tells no story about how we could come to have knowledge of these non-natural facts; it gives no mechanism through which they could be perceived. However, these objections may not be as powerful as they seem. Scanlon points out that as regards the problem of normative force, statements about reasons should not be seen as statements about the natural world; rather they can simply be regarded as statements that can be true or false, and therefore can be the object of belief.8 This is so because ‘statements about the reasons we have normally take the form of declarative sentences which obey normal laws of logic. If, then, we are disposed on reflection to confidently affirm judgements of these kinds we seem to need some reason not to take them as saying something which can be true and which can be the object of belief’ (Scanlon 1998:60). For example, the fact that it is currently raining can lead to the statement that the rain provides a reason for Dawn to take an umbrella with her to walk to the shop.


This statement can be true or false. Its truth is dependent on the objective conditions, that it is indeed raining and that Dawn is about to walk to the shop; and the subjective conditions, that Dawn does not enjoy being rained on (her subjective reaction to rain).9 If these conditions are fulfilled, then the statement will be true. It must also be noted that the statement can be true without this reason being an overriding reason for Dawn. She may have better reason to do something else, but this would not necessarily mean that the statement constitutes no reason for her, and is therefore false. This claim about the truth of reason statements resolves the problem of normative force for the belief account, because if the recognition of a reason is simply recognition of the truth of a reason statement as that statement applies to the agent, then this recognition will, according to Scanlon, have normative force of the necessary type. This also shows us how the metaphysical problem may be resolved. If the belief in question is not a belief about the natural world, then, Scanlon argues, it can simply be seen as a belief about reasons; and ‘if there are judgements about reasons which are naturally construed as declarative sentences and in which we have a high degree of substantive confidence, then why not conclude that these judgements do indeed say something true about their apparent subject matter’ (Scanlon 1998:61). The epistemology of reasons is also less problematic than it first appears. One important point here is that a belief about reasons is not an empirical belief about the natural world that is ‘outside’ us, and hence must in some way be represented to us. Reasons are not ‘outside’ us, as is the natural world, but according to Scanlon, they are in some sense independent of us. When we make a judgement about reasons, it does not seem as if we are recognising a fact that is independent of our judgement. But, if it is accepted that these judgements can be true or false, and that we can be mistaken about them, then it should also be accepted that these judgements concern a subject matter that is independent of us. Reasons are of course dependent on us in the sense that they apply to us. But, they are independent of us in the sense that we can recognise their existence without thinking that they apply to any agent in particular. To return to an earlier example, that a career in medicine is worthwhile is a reason to pursue such a career. This constitutes a reason in itself. The reason can then be said to apply to Naomi since she would enjoy a career in medicine, and it is only now that the reason can be seen as in any way dependent on, or related to Naomi. Scanlon uses the analogy of mathematics to explain the independence of reasons further. We can be right or wrong in mathematical reasoning, but there is nothing in the natural world that directly corresponds to mathematical objects. All that mathematics requires to be objective is that there are standards of correctness for mathematical reasoning, such that if these are not followed, then the conclusion is wrong (except where it is correct through chance). Thus, in order for judgements about reasons to be seen as concerning a subject matter independent of us, and therefore objective in the same way as judgements about mathematics, there must also be standards of correctness for reason judgements. While such


standards of correctness are not obvious (to say the least), they are more obvious in the case of reasons for belief than reasons for action. For example, an eyewitness account of a crime gives firmer grounds for belief about what happened than does hearsay. So, it appears plausible that we gain knowledge of reasons through reasoning in accordance with the relevant standards. The above consideration of the three objections to the belief account shows that this account can be seen as at least as plausible as the JSA account. Furthermore, Scanlon holds that whether the belief or the attitude account is accepted is of little consequence, as long as it is accepted that there are standards of correctness for reason judgements in either case, which implies that reason judgements can be true or false whether or not the actual judgements are beliefs or attitudes. Thus, the alleged existence of standards of correctness for reason judgements enables Scanlon to say that these judgements are objective, irrespective of whether they constitute beliefs or attitudes. It may be thought that Scanlon’s appeal to the appearance of reason judgements—the fact that they are declarative sentences in which we have a high degree of confidence—does not give enough support to the initial assumption that reason judgements can be true or false; and therefore that the argument is flawed. But, it must be remembered that the arguments of the previous section cast doubt on the presence of a subjective element (desire) in practical reasoning. If correct, this would suggest that practical reasoning is objective, and in this context, the assumption of truth value seems more plausible. Thus, for the present purposes it will be assumed that Scanlon has established that practical reasoning does have standards of correctness, and that it is therefore plausible to hold that reason judgements can be true or false. Scanlon then gives an answer to the question of how we can have knowledge of the reasons we have. This involves looking at the grounds we have for making the decision that X is a reason for attitude A, and the grounds that we have for thinking that reason judgements can have truth value. He first provides a fourstage process for deciding whether X may be a reason for A:10 1. X seems to me to be (or not to be) a reason for attitude A. 2. I may decide whether or not X really is a reason for A. 3. I decide whether, taking account of X and all other relevant reasons, there is sufficient reason for adopting A. 4. I may come to have attitude A. This procedure is meant to take us from the position of a consideration seeming to be a reason, to a justified decision that that consideration really is, or is not, a reason for the attitude in question. The decision is justified by the critical stages (2 and 3), because these stages give us a clearer and more detailed idea of the reason in question, and should remove or at least lessen the influence of distorting factors, such as anger, or desire from our deliberation. That such factors as rage are seen as distorting influences here is a substantive judgement, which is


generally agreed upon, and forms part of a background of stable, generally agreed judgements against which particular reason judgements are made. This suggests a holistic approach to reason judgements, wherein all other judgements are held fixed and comprise points of reference for the judgement in question.11 How do we know which considerations constitute reasons? Since reasons are requested, or offered, for judgement-sensitive attitudes, then it appears that the content of the judgement-sensitive attitude in question should provide guidance in identifying the kind of considerations that could count in favour of it. If the content of the attitude does not provide such guidance, then the question of whether or not a consideration is a reason for it, according to Scanlon, will make no sense. Background judgements about the attitude will also help to clarify whether the consideration is a reason, for example, it is rational for Nick to fear the dog that is trying to bite him, because this represents a serious threat of physical harm, but it is not rational for Nick to fear the dog that is wagging its tail and making no attempt to come near him. In the former case, the dog’s actions constitute a valid reason for Nick to adopt the attitude of fear, but in the latter, the dog’s actions do not constitute a valid reason, due to background judgements concerning, among other things, the strength of the threat of physical harm. It is also possible during the course of deliberation for background judgements to be modified, if this seems appropriate. For example, deliberation about one potential reason may highlight an inconsistency between beliefs that must be resolved, leading to a change in a background judgement. Scanlon argues that in deciding whether or not a consideration is a reason, what we are doing is deciding whether this reason would be included in the most complete and coherent account of what reasons there are. So, if the agent decides it would be included, then she is pronouncing it to be a valid reason. Beliefs about reasons that are the conclusions of detailed deliberation in accordance with the four stages above, should, in general, be very stable, and Scanlon also notes that there is a significant degree of interpersonal agreement regarding reasons that appear to be stable under this kind of reflection. He concludes that this ‘provides ample ground for saying that judgements about reasons for action are the kind of things that can be correct or incorrect’ (Scanlon 1998:70). The final part of Scanlon’s theory of reasons is his account of the importance of other people’s reasons and specifically the principle of the universality of reason judgements. The point here is that any judgement about our own reasons entails claims about what reasons others have, or would have under relevantly similar circumstances. So, if Ryan takes set of factors G as giving him reason to do A in circumstances C, then he is committed to thinking that any other agent in C, facing factors G, also has reason to do A. This is not a claim about what considerations count as reasons, as G is entirely open. The principle of ‘the universality of reason judgements is a formal consequence of the fact that taking something to be a reason for acting is not a mere pro-attitude toward some action, but rather a judgement that takes certain considerations as sufficient grounds for its conclusion’ (Scanlon 1998:74). The principle also reflects the implication


throughout this account of reasons that reasons exist, at least potentially, independently of any agent. So, if an agent should happen to be in a particular situation, or have a particular subjective reaction, then reason X automatically becomes applicable to that agent. As highlighted above, it is circumstances, stable background judgements, and in some cases subjective reactions that are the determining factors for the existence and applicability of reasons, any other facts about the agent, such as his desires, are irrelevant. This principle of the universality of reason judgements is the final piece of Scanlon’s argument for the conclusion that moral requirements are not escapable. The previous arguments in this section established the objectivity of reason judgements, standards of correctness for reasoning to reach these judgements, and their truth value. It was then argued that there is a procedure by which we can (usually) reach correct judgements about reasons, and the addition of the principle of the universality of reason judgements allows us to make correct judgements about other people’s reasons, and them to make correct judgements about our reasons. Thus, Scanlon’s theory of reasons can be seen as providing a detailed argument in favour of the conclusion that moral requirements, through providing us with reasons to hold various attitudes, and act in accordance with those attitudes, are inescapable. Agents may not recognise moral reasons, but this would be a mistake. Scanlon’s theory implies that these reasons apply to them whether or not the reasons are recognised. In his theory of reasons, Scanlon aimed to establish that reason judgements are objective in the sense that they can be considered right or wrong in the context of the information available to the agent (agents are reasonable, not ideally rational). To this end, he provided a negative argument—that there is little or no subjective component in reason judgements, since desire plays no motivational or justificatory role—and a positive argument, which claimed standards of correctness and truth value for reason judgements. The main weakness of his account is in his argument against desire playing a motivational role. I suggest that further exploration of this issue may yield substantive criticism of the argument against there being a subjective component in practical reasoning, which may also affect the positive argument for the objectivity of reason judgements. ACKNOWLEDGEMENTS I am grateful to the Goodman Trustees for providing financial backing for my PhD. I would also like to thank Sue Mendus and Matt Matravers for their continuous support and encouragement.


NOTES 1. This is almost certainly one of the reasons for the shift from a desire-based to a reason-based contractualism. 2. The accessibility of information is an important factor in determining reasonableness. It seems that the more important the issue, the greater the effort an agent can reasonably be expected to make to gain access to the relevant information. 3. However, it is important to reiterate that Scanlon has not always held this view. In his earlier work on contractualism (Scanlon 1982) desire played a far greater role than it does in the work currently under consideration (Scanlon 1998). 4. For more on akrasia of belief and self-deception, see Hurley 1989:131–3. 5. Adapted from Scanlon 1998:37. 6. This emphasises the fact that DA desire, at least, cannot be a JSA, because a central feature of DA desire is that it is not judgement sensitive. 7. Scanlon admits that such subjective reactions may occasionally operate as sources of reasons, but argues that they are ‘rather trivial cases, not central examples’ and so should not be used as the basis for a model of the source of reasons (Scanlon 1998:48). 8. The overall tone of the theory of reasons (see especially Scanlon 1998:32–3, 70–72) suggests that we are to assume that agents making reason judgements are reasonable rather than ideally rational, which means that reason judgements should not be seen as right or wrong in an absolute sense, but rather right or wrong in the context of the available information. 9. This reason clearly has both objective and subjective truth conditions, but it must be remembered that moral and prudential reasons are thought by Scanlon not to be dependent on agents’ desires in any way, and therefore such reasons only have objective conditions. 10. The following is adapted from Scanlon 1998:65–6. 11. Although Scanlon himself does not say that this procedure is holistic.

REFERENCES Hurley, S.L. 1989. Natural Reasons. Oxford: Oxford University Press. Nagel, T. 1970. The Possibility of Altruism. Oxford: Clarendon Press. Rachels, J., ed. 1998. Ethical Theory 2: Theories About How We Should Live. Oxford: Oxford University Press. Scanlon, T.M. 1982. ‘Contractualism and utilitarianism’. Reprinted in Rachels 1998. 1998. What We Owe to Each Other. Cambridge, MA: Harvard University Press.

3 The Magic in the Pronoun ‘My’ SUSAN MENDUS

In Chapter 4 of What We Owe to Each Other T.M.Scanlon writes: The fact that an action would be wrong constitutes sufficient reason not to do it (almost?) no matter what other considerations there might be in its favor. If there are circumstances in which an agent could have sufficient reason to do something that he or she knew to be wrong, these are at best very rare. But if right and wrong always or even almost always take precedence over other values, this is something that requires explanation. How can it make sense, if we recognize values other than right and wrong and take them seriously, to claim that reasons of this one kind have priority over all the rest? I will refer to this as the problem of the priority of right and wrong over other values. (Scanlon 1998:148) And he goes on to note that this problem, the problem of the priority of right and wrong over other values, is not simply a matter of practical conflict. It runs deeper than that, since it is possible not only that morality may sometimes dictate an action different from and in conflict with the action required by another value, but also that the reasons offered by morality may conflict with reasons offered by other values. Here Scanlon appeals to Bernard Williams’s example of a man faced with a choice between saving his wife and saving a stranger. He notes Williams’s suggestion that it would be contrary to the impartial regard for every person that these forms of morality [impartialist forms] require to save one of them because of some special tie, for example because that person was your friend or your spouse. This is not merely a case of practical conflict. The suggestion is that a morality that required this kind of impartiality (that declared love or friendship to be an impermissible reason for saving one person rather than the other) would rule out love and friendship altogether, since it is essential to these relations that one would have reason to give preference to a friend or loved one in such a situation. (Scanlon 1998:160)


The problem, then, is this: we need to know whether impartialist morality leaves room for other values, such as the value of friendship, and, if so, how it defends its own claim to priority in cases of conflict. As the second quotation demonstrates, Scanlon is aware of the fact that some philosophers use the possibility of conflict to express their dissatisfaction with impartialism quite generally. For Williams, impartialism’s perceived inability to accommodate values of love and friendship renders it unacceptable as a moral theory. However, even if impartialist theories can make room for values such as love and friendship, they still need to defend and justify the priority claim identified in the first quotation. There are, therefore, two distinct questions: first, ‘does impartialist morality drive out other values, such as the values of love and friendship?’; second, ‘even if it does not, how can the reasons of impartial morality have priority over reasons grounded in other values?’ In what follows, I aim to address these questions, and to do so through a critique of Scanlon’s account. My claim will be that Scanlon identifies important truths about the nature of values such as love and friendship, but that ultimately he misrepresents the relationship between those values and impartial morality. In particular, his conclusion that ‘compatibility with the demands of interpersonal morality is built into the value of friendship itself’ (Scanlon 1998:165) reverses the correct order of explanation by urging us to see friendship as grounded in morality, when in fact morality is, to a significant extent, grounded in friendship. My argument to this conclusion proceeds as follows: first, I consider one impartialist (William Godwin) who is happy to accept that impartialism drives out other values, such as love and friendship. However, Godwin’s thoroughgoing impartialism serves merely to highlight the fact that, in general, impartialists accept the need to demonstrate the compatibility between values of right and wrong (impartially understood) and other values. Since thorough-going impartialism results in absurdity, the real challenge is to show not whether, but how, impartialism can leave room for other values. I therefore consider one very important strategy for attaining this end: the strategy of distinguishing between two levels of impartiality. I conclude that, while this strategy avoids the absurdities of thorough-going ‘Godwinian’ impartialism, it is nonetheless incomplete because it fails to show why, in cases of conflict, moral values should (almost) always be accorded priority. Scanlon’s second question still stands in need of an answer. I then discuss two answers to that question: the answer offered by Scanlon himself in What We Owe to Each Other, and the answer offered by Christine Korsgaard in Sources of Normativity (Korsgaard 1996). While finding each of these problematic, I suggest ways in which elements from each might be combined to provide a third, and more satisfactory, answer. However, that third answer is one that requires a reversal of the relationship Scanlon believes to hold between morality and other values. It requires, not that the value of friendship be seen as grounded in morality, but rather that the value of morality be seen as grounded in friendship. I begin, however, with that most robust of impartialist thinkers, William Godwin.

THE MAGIC IN THE PRONOUN ‘MY’ 29

Godwin and the Famous Fire Cause In a now notorious passage in Political Justice Godwin asks what should be done in a case where two people are trapped in a burning building and only one of the two can be saved. His example supposes (depending on which version you read) that the two people are the Archbishop Fenelon and his valet. Of this case, Godwin writes: The illustrious archbishop of Cambray was of more worth than his valet, and there are few of us that would hesitate to pronounce, if his palace were in flames, and the life of only one of them could be preserved, which of the two ought to be preferred…suppose I had myself been the valet, I ought to have chosen to die rather than that Fenelon should have died… Suppose the valet had been my brother, my father or my benefactor. That would not alter the truth of the proposition. The life of Fenelon would still be more valuable than that of the valet; and justice, pure unadulterated justice, would still have preferred that which was most valuable…what magic is there in the pronoun my to overturn the decisions of everlasting truth? My brother or my father may be a fool, or a profligate, malicious, lying or dishonest. If they be, of what consequence is it that they are mine? (Godwin 1976:70) It is significant that this robust defence of impartiality was roundly denounced by Godwin’s contemporaries: Charles Lamb referred to it mischievously as a case in which Godwin appeared as ‘counsel for the Archbishop Fenelon against my own mother’, while others took the example to be proof positive that Godwin was a ‘desiccated calculating machine, devoid of normal human feelings, his great head full of cold brains’ (Locke 1980). Even in his own day, then, Godwin’s sanguine acceptance that impartial morality leaves no room for values of love and friendship was thought to be at best comical, and at worst positively inhuman. Nor has his reputation improved with the passing of the years. Lawrence Becker refers to the ‘evident foolishness’ of the belief that we should act with perfect Godwinian impartiality in all aspects of our lives (Becker 1991), and Brian Barry has concluded that, if Godwin had not existed, anti-impartialists would have to have invented him (Barry 1995). In short, it is, and always has been, generally acknowledged that if impartialism is incapable of accommodating values of love and friendship, then it is doomed. On pain of absurdity, impartialists must show that their moral theory does not drive out other values entirely. How, then, is this to be done? Two Levels of Impartiality One very popular solution involves distinguishing between two levels of impartiality: impartiality at the level of day-to-day decisions, and impartiality at


the level of principle selection. Godwin’s mistake, it is argued, lies in his failure to notice the difference between these two levels and between the distinct questions associated with each: the question ‘What ought I to do?’, on the one hand, and the question ‘What morality ought there to be?’, on the other. Impartialism is an appropriate answer to the second question, but not thereby to the first, for we can (and should) insist that moral principles be impartial without being committed to the belief that each and every decision be governed by impartialist considerations. This distinction between impartiality at the level of principle selection and impartiality at the level of day-to-day action has been widely embraced as a way of adhering to an impartialist moral theory while escaping the absurdities of Godwin’s conclusion. Thus, Brian Barry writes: ‘What makes Godwin the archetypal exponent of first order impartiality is not his idea that everyone ought to do the best thing open to him. What makes him distinctive is his idea that the best thing to do is to behave with strict impartiality always and that this entails giving no weight to prior commitments or personal attachments’ (in Kelly 1998: 246). Impartialism, then, is an attempt to answer the question ‘What morality should there be?’ and we can readily concede that the principles of morality should be impartial while also denying that those principles should be so strict as to leave no room for prior commitments or other values. On the contrary, and as Barry insists, any sensible form of impartialism will adopt principles that allow ample room for discretion in particular cases. A similar solution is proposed by Marcia Baron, when she writes: ‘Critics suppose that impartialists insisting on impartiality at the level of rules or principles are committed to insisting on impartiality at the level of deciding what to do in one’s day-to-day activities’, but this is not so, and once the two-level distinction is noted we will see that ‘there is far less and far less deep disagreement between partialists and impartialists than is often supposed’ (Baron 1991:842, 857). So, by adopting a two-level account which distinguishes between principle selection and everyday action, we can demonstrate the ability of impartialist moral theory to accommodate values such as love and friendship. The answer to Scanlon’s first question (does impartialism drive out other values?) appears to be ‘It does not’. However, insofar as impartialism permits other values, it must explain and defend its own claim to priority in cases where conflict between values occurs. The very ability of impartialism to answer the first of Scanlon’s questions itself gives rise to his second question: ‘How can the reasons of impartial morality have priority over reasons grounded in other values?’, for either impartialism allows other values, or it does not. If it does not, then it seems to be committed to the absurdity of Godwin’s conclusion. If it does, then it must explain what relationship holds between the values of impartial morality and other values. In particular, it must explain why the reasons of impartial morality should have priority. It is to this problem—the priority problem—that I now turn.


Refining the Priority Question I began by quoting Scanlon’s claim that ‘if right and wrong always or even almost always take precedence over other values, this is something that requires explanation. How can it make sense, if we recognize values other than right and wrong, and take them seriously, to claim that reasons of this one kind have priority over all the rest?’ (Scanlon 1998:148). This, as we have seen, is the priority question, and, in order to explain and assess it more precisely, I shall draw upon an example from Marcia Baron. Baron envisages a case in which my son is ill and in desperate need of medical treatment. There is a long waiting list for treatment and every chance that, by the time my son’s turn comes up, it will be too late. However, I am in a position to pull strings and secure early treatment for him. Commenting on the case, Baron remarks: ‘This much, I think is clear: we expect people to want very badly to pull strings in such a situation, and we think well of them for being tempted to do so. We would think less highly of someone who felt no such tug at all, and who immediately and easily ruled out that course of action, recognising that it would be unfair. Yet we see the unfairness, and for this reason we want rules to prevent, as far as possible, people from doing this thing that they are tempted to do’ (Baron 1991:855–6). I choose this example because there are several respects in which it highlights the precise nature of the priority question: first, it exemplifies the central fact to which Scanlon draws our attention, which is that while we do indeed take seriously values other than the values of right and wrong (in this case, the value of the mother’s love for her child), we nonetheless feel that, if impartialist considerations apply at all, then they should take precedence over other considerations. Second, the example invites us to consider things from the point of view of the mother. We need to know, not simply what impartialist thinking requires, but why the mother should be moved by impartialist thinking, rather than being moved by the needs of her child. Finally, the example draws our attention to the possibility that we might sometimes ‘think better’ of someone who is at least tempted to go against the dictates of impartial morality and act on other values. In short, the example prompts us to ask the priority question: ‘Why should I give priority to what impartial morality dictates rather than giving priority to the needs of my child?’, where that question is posed from the agent’s perspective, and where it concerns the status of the reasons offered by impartial morality vis-à-vis the reasons offered by other considerations, including considerations of partiality. That this is the precise nature of the priority question is clear from Scanlon’s insistence that an acceptable moral theory must show how considerations of right and wrong (impartial considerations) can play a certain role in the thinking of the agent (Scanlon 1998:149). Nor is he alone in taking this to be the challenge posed. In Sources of Normativity Christine Korsgaard also insists that ‘a theory that could explain why someone does the right thing—in a way that is adequate


from a third-person perspective—could nevertheless fail to justify the action from the agent’s own, first-person perspective…the question is should we allow ourselves to be moved by the motives which morality provides?’ (Korsgaard 1996:14). The priority question is therefore a question posed from the point of view of the agent. Additionally, the priority question concerns the compatibility (or otherwise) of the reasons offered by impartial morality with the reasons offered by particular loyalties or other values. Thus (Scanlon again): ‘Values can conflict in a practical sense when they give rise to incompatible demands to action, but even when no act is in view they can conflict in a deeper sense when one value involves giving certain considerations a status as reasons that another value rules out’ (Scanlon 1998:13). The defence of priority must therefore address the agent him/herself and it must show whether and why the reasons of morality, impartially understood, should take priority over reasons grounded in other values, such as the value of friendship. These two features of the problem anticipate two responses to it, and I will now discuss those responses in an attempt to clarify the overall debate. The Reductivist Response This response argues that, although there may appear to be conflict between the reasons of morality and the reasons offered by other values (such as friendship), in fact the appearance is illusory because the reasons offered by other values are themselves grounded in reasons of morality. This is, in fact, the strategy adopted by Scanlon, and it is evident in the following quotation: the conception of friendship that we understand and have reason to value involves recognising the moral value of friends qua persons, hence the moral claims of nonfriends as well. No sacrifice of friendship is involved when I refuse to violate the rights of strangers in order to help my friend. Compatibility with the demands of interpersonal morality is built into the value of friendship itself (Scanlon 1998:165, my emphasis). Scanlon’s argument for this conclusion has two component parts. First, he notes that real friendship, at least as we understand it, requires a recognition of the friend as a separate person in his or her own right. So the moral requirement of respect for persons does not stand opposed to, but is a presupposition of, friendship properly understood. Second, he claims that, in the absence of this recognition, we would not have friendship at all because ‘the moral standing [of the “friend”] would be too dependent on the contingent fact of affection’ (Scanlon 1998:164). The value of friendship, then, is grounded in the value of morality and, if it were not so grounded it would cease to be friendship as we know it. Scanlon’s response is reductivist in the sense that it construes reasons of friendship as compatible with, because grounded in, reasons of morality. And it


further construes morality as necessary for differentiating between real friendship and contingent affection. The Identity Response The second response, which I shall call the ‘identity response’, is offered by Christine Korsgaard, when she writes: Personal relationships, then, as a form of practical identity, are independent sources of obligation, like moral obligations in their structure, but not completely subsumed under them. And the thought of oneself as a certain person’s friend, or lover or parent or child can be a particularly deep form of practical identity. There is no obvious reason why your relationship to humanity at large should always matter more to you than your relationship to some particular person; no general reason why the laws of the Kingdom of Ends should have more force that the laws of the Kingdom of Two. I believe this is why personal relationships can be the source of some particularly intractable conflicts with morality. (Korsgaard 1996:128) Again, there are two features of the account which should be noted. The first is that, unlike Scanlon, Korsgaard sees the obligations of friendship and the obligations of morality as having distinct, though analogous, sources. It is not, for her, true that the requirements of friendship are grounded in the requirements of morality and, for this very reason, the possibility of conflict is permanent and persistent. Second, Korsgaard sees the demands of friendship as involving one’s practical identity. To be required to reject my obligations to my mother, my husband, my friend, or my child in the name of morality is (at the limit) to jeopardise my understanding of who I am. So this response affirms the persistence and intractability of conflicts between morality and other values. It sees such conflicts as ones in which one’s practical identity may be at stake, and it offers no general hope of resolution. So far, then, I have identified the following responses to the priority problem: 1. Conflicts between morality and friendship are more apparent than real because, on inspection, reasons of friendship turn out to be grounded in morality. Without that grounding, friendship becomes simply a matter of contingent affection. 2. Conflicts between morality and friendship are both real and intractable. They are real because the obligations of morality and the obligations of friendship have distinct sources. They are intractable because both sources involve our practical identity and there is no reason in principle why that part of my


practical identity which flows from being moral should take precedence over that part which flows from deep personal relationships and commitments. However, the difficulty with both these responses is that they seem to underscore, rather than answer, the problem with which we began, which was to explain the priority of the moral in cases where its reasons conflict with the reasons offered by other values. The reductivist response denies that there is (genuine) conflict, while the identity response denies that morality has (must have) priority. To progress further, we need to see both what is wrong with each account, and what truth lies within each. I shall therefore provide a bit more detail about what exactly I think is true in each account, then I shall say something about what seems to be mistaken in each, and finally I shall attempt to reconcile the truths as a way of constructing a better understanding of the priority problem in general and of the relationship between morality and friendship in particular. The Truth in Reductivism Scanlon’s resolution of the priority problem involves the claim that reasons of friendship are grounded in, and therefore are compatible with, reasons of morality. Being a true friend involves ‘recognizing the friend as a separate person with moral standing—as someone to whom justification is owed in his or her own right, not merely in virtue of being a friend’ (Scanlon 1998:164). Indeed, he goes further, claiming that relationships of friendship depend upon impartial considerations and that without them friendship disintegrates into mere contingent affection—which is not friendship at all. He then goes on to illustrate his claim via the following (rather puzzling) example: There would be something unnerving about a ‘friend’ who would steal a kidney for you if you needed one. This is not just because you would feel guilty towards the person whose kidney was stolen, but because of what it implies about the ‘friend’s’ view of your right to your own body parts: he wouldn’t steal them, but that is only because he happens to like you. (Scanlon 1998:164–5) In assessing this example it is crucial to bear in mind the fact that Scanlon’s argument is not addressed to the problem of practical conflict, but to the question of whether the reasons of morality conflict with the reasons of friendship. He concludes that they do not, because friendship itself is a moralized notion such that, even within close friendships, impartial considerations, such as respect for the friend as a separate person with rights of her own, have application. This point seems to me to be both true and important. Although our relationships with friends, family and loved ones may appear very different from our relationships with strangers, there are dangers inherent in over-drawing the distinction or


adopting an overly romantic understanding of the nature of close personal relationships. To see why this is so, consider the case of Nora in Ibsen’s play A Doll’s House (Ibsen 1965). The two central characters of the play are Torvald Helmer and his wife Nora. Torvald is a successful businessman who, as the play opens, is on the verge of promotion to a high-ranking position in the local bank. The promotion will bring with it wealth and status within the local community. It represents the culmination of Torvald’s ambition. However, success and status have been hard won, and during the early years of his marriage to Nora he suffered serious ill health. At that time, and unknown to him, Nora borrowed money to pay for his medical treatment. Because women were not allowed to borrow money without a male guarantor, she forged her father’s signature on the official loan document. As the play progresses it becomes increasingly likely that this misdeed will be discovered, Torvald’s promotion will be jeopardised, and his reputation as a pillar of the community will be destroyed. In the final scene Torvald does indeed discover what Nora has done. She hopes that the discovery will make him realise how much she has loved him and how much she has been prepared to sacrifice for him. To her horror, he receives the news as proof that he has been married to a liar, a forger and a cheat. His wife is not the innocent and guileless ‘doll’ he supposed her to be, but a common criminal. The final scene focuses on the discussion which ensues and in which their different and conflicting conceptions of what is important in life are exposed. Throughout the play, we are presented with a vision of Nora as a bird, a plaything, a creature of light and joy, a capricious, whimsical, unpredictable child. This is the way her husband, Torvald, perceives her and this perception informs their relationship. In the end, it is their undoing. When, in the final scene, Nora announces that she will leave Torvald and their children, he reminds her of her duties to the children and to him as her husband. She replies that she has ‘more sacred’ duties—duties to herself. And this reply gives voice to her deep resentment that Torvald’s ‘love’ for her has never been informed by a recognition of her as a separate and autonomous person, deserving of respect. Indeed, what she now sees is that this never was love at all—merely contingent affection for the ‘doll’ Torvald believed her to be. And so she concludes that they never had a ‘true marriage’. The case of Nora exemplifies what I take to be an important truth in Scanlon’s account: even the most personal and intimate of relationships—the relationship of marriage—requires that each party recognise the other as a person in his or her own right. In the absence of this recognition, there are reasons for doubting whether the relationship genuinely is a relationship of friendship, or of love. Ibsen proposes what Scanlon confirms, that ‘the conception of friendship we understand and have reason to value involves recognising the moral value of friends qua persons’ (Scanlon 1998:165). The truth in the reductivist response is that friendship, if it is to be real friendship, must be informed by considerations


of impartial morality, and for that reason cannot be completely in conflict with it at the level of reasons. The Truth in Identity What, then, is the truth in the identity response? Korsgaard gives the following account of what is central to and distinctive of personal relationships (including relationships of love and friendship): A personal relationship is a reciprocal commitment on the part of two people to take one another’s views, interests and wishes into account. This kind of reciprocity leads to what Kant called ‘a unity of will’, for the two parties must, at least in the areas their relationship is concerned with, deliberate as one. Personal relationships are therefore constitutive of one’s practical identity. (Korsgaard 1996:127) What is significant here is the identification of something beyond contingent affection which is partly definitive of personal relationships. Where Scanlon concludes that, in the absence of a moral grounding, personal relationships will become merely a matter of contingent liking and disliking, Korsgaard identifies another possibility—reciprocal commitment. This goes beyond contingent affection, since it is constitutive of one’s practical identity, but it is not (she says) grounded in morality. On the contrary, it has a distinct source and may therefore conflict with morality. The concept of reciprocal commitment as constitutive of one’s practical identity is (for me at least) somewhat elusive, so let me try to articulate what I take it to imply. Consider, again, the case of Nora in Doll’s House. In discussing the truth in Scanlon’s account, I suggested that one source of the tragedy is Torvald’s inability (or refusal) to recognise Nora as a separate person with moral standing—as someone to whom justification is owed in her own right, and not simply as a wife and mother. It is this which prompts Nora to conclude that Torvald has never really loved her and that their marriage has never been a true marriage. However, there is another dimension to Nora’s conclusion. This is her realisation that she and Torvald have never been reciprocally committed in the sense of having shared ideals. At the point in the play when Torvald realises that his honour will be impugned if Nora’s crime becomes public, he rejects her in words which make plain that his projects and values as an individual have not been modified, or shaped, by hers. His own ideal of the supremacy of honour, the law and public affirmation stand untouched by her ideals of loyalty to her husband and protection of her children before all else. Thus, when Torvald declares that it is unreasonable to expect him to sacrifice his honour for her (‘No man would do that’, he says), Nora concludes that throughout their marriage they have been ‘strangers’ to one another, and I take it that part of what is implied by


this conclusion is that they have never been reciprocally committed to one another. He has never gone beyond seeing his projects as exclusively his, never had an understanding of their relationship as involving ideals which are theirs. Taken together, the two facets of Doll’s House highlight the truths in Scanlon’s and Korsgaard’s accounts: they exemplify both the fact that friendship is a partly moralised concept which requires acknowledgement of the friend as a separate person, and the fact that friendship involves (or may involve) a sharing of interests such that the projects of the other person become one’s own. However, the acknowledgement of these truths also serves to highlight the flaws in their responses, and I shall now discuss these as a preliminary to the construction of an alternative account of the priority of moral reasons. The Flaws in Reductivism One difficulty with Scanlon’s account is that, as noted above, it appears to dissolve, rather than resolve, conflicts between friendship and morality. It appears to provide a response to cases of conflict which deny its very possibility. For Scanlon, this is a serious flaw, since his aim is to provide a moral theory which is phenomenologically accurate, and he makes much of the fact that his account is just that (Scanlon 1998:155, 158, 187). In this case, however, it is not clear that the theory has the virtue claimed for it: the kidney theft example just seems too easy, for from the fact that I would not steal a kidney in order to help my friend it does not follow that compatibility with the demands of impartial morality quite generally is built into the value of friendship, even if we restrict our discussion to the distinctive area of what we owe to each other. Presumably, there is more to what we owe to each other than simply the non-violation of rights, and if we turn from the kidney theft case to other cases, such as the one cited by Baron, we find that (intuitively, at least) what we owe to strangers may have far less motivational force. Indeed, what is most problematic about this account is that it is not clear what motivational purchase it will have on the agent to whom it is addressed. Recall that the priority problem requires us to show that reasons of morality are compatible with reasons of friendship, and to do that in a way which addresses the agent him or herself. In the more difficult cases of conflict, it is far from clear that the reductivist approach can respond to that demand. The agent may concede that friendship will not be genuine unless it acknowledges the claims of impartial morality, yet still wonder why those claims should matter more than the claims of friendship. However, even if this phenomenological point is moot, there is a second difficulty with the reductivist strategy, which is that, in the strong form in which Scanlon embraces it, it requires a highly moralised understanding of friendship, one which seems to end in the assertion that cases in which morality is rejected in favour of friendship are not cases of genuine friendship at all, but merely cases of contingent affection. In short, Scanlon’s story is one according to which


friendship just is contingent affection, suitably moralised. And this, for reasons which I hope will become clear in a moment, seems wrong. In the first place, then, the reductivist response seems unable to provide the requisite motivational purchase on the agent—at least in the most difficult cases. And in the second place, the reductivist response succeeds only by moving too rapidly from the claim that real friendship has a moral component to the conclusion that any act of friendship not grounded in reasons of morality must be grounded in contingent affection and therefore not be a case of true friendship at all. This, however, seems too high a price to pay for the resolution of conflict, and a price which distorts our ordinary understanding of friendship, for even if it is, in general, true that reasons of friendship acknowledge reasons of morality, it does not follow that this consideration will (or should) necessarily weigh with the agent who is tempted to go against the thing which morality requires in order to help a friend. Nor is it true, as Scanlon claims, that such a person will cease to be a genuine friend should he or she refuse to do the thing required by morality, and to say otherwise is to deny the distinction between bad acts done from friendship and bad friendships. I take it that it is this distinction which Baron’s example highlights: it is arguable that the mother who pulls strings for her son and secures early medical treatment for him does a bad act, but it is not clear that she is thereby a bad mother, for while it is possible to be a bad mother, or a bad friend, by ignoring the dictates of morality, Scanlon’s account rests on the contention that all failures of morality are thereby failures of friendship, and this seems simply wrong. The Flaws in the Identity Response However, if the reductivist response (Scanlon’s response) is incapable of generating motivational purchase on the agent, the identity response (Korsgaard’s response) fails to generate the right kind of motivational purchase for the resolution of the priority problem. It explains the agent’s desire to favour a friend by reference to the significance which the friendship has for the practical identity of the agent himself (or herself), and this can appear to be tantamount to self-indulgence, or even narcissism, for a decision to help one’s friend ought, surely, to flow from a concern for the friend, not from a concern for oneself and one’s own practical identity. Additionally, it is a large claim to suggest that cases in which I fail to respond to an obligation (whether of friendship or of morality) are cases in which my identity comes under threat. Most often, they are simply cases in which I feel badly about myself. And even if they are cases in which my identity is threatened, it is not clear why my identity should matter more than anything else. The appeal to identity thus seems both exaggerated and ignoble. The reductivist and identity responses both answer to an important dimension of the priority question: the reductivist response shows why reasons of friendship are not entirely divorced from reasons of morality. However, in Scanlon’s case,


it does so in a way which dissolves the dilemma entirely and fails to account for the agent’s motivational difficulty. By contrast, the identity response answers to the agent’s motivational difficulty, but does so in a way which exaggerates what is at stake in conflicts of this kind, and in any case provides the wrong kind of motivation—it provides a motivation which looks to the interests of the agent herself, and this appears to justify what is (or may be) merely self-indulgent. To recap, an acceptable response to the priority question must do two things: it must provide reasons which explain the priority of moral values over other values, and it must provide reasons of a sort which will gain purchase on the agent’s own motivations. By showing how the requirements of friendship contain within them an acknowledgement of the moral, the reductivist account satisfies the first requirement, but not the second. By showing how the claims of friendship may involve more than merely contingent affection, the identity account satisfies the second requirement but not the first. The reductivist account gives reasons to do what morality requires, but these reasons are of a sort which will struggle (in hard cases) to gain motivational purchase on the agent, and which will ultimately deny that those cases are indeed hard. In this respect, it will fail to do justice to the way the agent sees the problem (it is phenomenologically defective). On the other hand, the identity account provides a motivational story, but one which makes conflict not merely difficult, but completely intractable. Moreover, the motivational story it tells is one which construes, and indeed defends, reluctance to do what is morally required by appeal to the importance of one’s own moral purity. If the abandonment of moral requirements is to be defended it ought not, I think, to be on those grounds. What, then, is the way forward? A Rapprochement In discussing the identity account, I endorsed its assertion that friendship is more than contingent affection suitably moralised. It also involves reciprocal commitment on the part of the friends, a sharing of interests between them such that the interests of the friend become one’s own interests too. And I was critical of Scanlon’s neglect of this dimension of friendship—of his silence about the ways in which my commitment to my friends may extend beyond mere contingent affection. But if he is neglectful of this as an essential component of friendship, he is insistent on it at as a presupposition of morality. He writes: ‘Unlike friendship, morality is commonly seen as a form of constraint, not as a source of joy or pleasure in our lives.’ However, he continues: ‘I am suggesting that when we look carefully at the sense of loss occasioned by charges of injustice and immorality, we see it as reflecting our awareness of the importance for us of being in unity with our fellow creatures’ (Scanlon 1998:163). For Scanlon, this need is one of the things that invests our lives with meaning and indeed constitutes a source of our practical identity. So the claim that friendship


involves contingent affection, suitably moralised, must be read in the context of a conception of morality as something which can be a source of joy insofar as it represents a non-instrumental or expressive concern to be at one with others. Moreover, this consideration is widely invoked by impartialists (more widely than is usually admitted, I think). David Gauthier notes that ‘morality takes on a different coloration when viewed in relation to participation. For asocial seekers and strivers morality could be no more than an unwelcome constraint. But for those who value participation, a morality of agreement, although still a source of constraint, makes their shared activity mutually welcome and so stable’ (Gauthier 1986:337). The admission ticket to morality proper is one which is not available to asocial seekers and strivers but (again) is reserved for those who welcome shared activity and the mutuality it implies. Similarly Rawls claims that although moral feelings are ‘admittedly unpleasant, in some extended sense of unpleasant, there is no way for us to avoid a liability to them without disfiguring ourselves’. And, he continues, ‘this liability is the price of love and trust, of friendship and affection’ (Rawls 1971:489). In all these cases, mutual recognition, a concern to be at one with others, ties of affection—what we might, in very general terms, refer to as non-instrumental, or expressive, concern for others—are deemed to be the preconditions of being genuinely and fully moral. They are the terms on which alone we can find morality anything other than a necessary evil. Indeed, and for Rawls, they are the terms on which we can retain our humanity. However, if all this is so, then we can perhaps see more clearly what is troubling about the troubling cases in which the priority question arises. These are cases in which those non-instrumental attitudes towards others, which formed the necessary condition of entry into morality, are undermined by the demands of morality itself. Pace Scanlon, they are not simply cases in which morality is contingently and momentarily set against affection. Pace Korsgaard, they are not cases in which morality is intractably and fundamentally set against practical identity. They are, in a sense, cases in which morality is set against itself. To explain this a bit more fully, I will return (briefly) to the truths identified in each of the reductivist and identity responses. There I suggested that the truth in the identity response lies in its recognition of reciprocal commitment: friendship implies that I see my own interests as being tied up with the interests of my friend. In the strongest cases, this will mean a sharing of interests such that the distinction between what is in her interest and what is in my interest becomes blurred. But even in weaker cases of friendship, it will imply a non-instrumental attitude towards the interests and concerns of one’s friend. However, I also suggested that the truth in the reductivist response lies in its recognition that my friend is, nonetheless, a separate person, and deserving of respect as such. Scanlon’s appeal to mutual recognition, to the importance to us of being in unity with our fellow creatures, is thus an appeal, not to one truth, but to two, for mutual recognition implies both the importance of treating others as persons in their own right (recognising that they are people first and friends, lovers, partners


thereafter), and the possibility of coming to see their concerns as our own, or at least of valuing those concerns non-instrumentally. But if it is also true, as it seems to be for Scanlon, Gauthier and Rawls, that something like mutual recognition in this second sense is the motivational underpinning of impartial morality, then the situation in the troubling cases is that here we seem to be required to abandon, in the name of morality, that concern for others which enabled us to become properly moral in the first place. To put the point bluntly, where Scanlon’s explicit argument is that the claims of friendship are grounded in morality, his implicit position is that the claims of morality are grounded, if not in friendship, then at least in non-instrumental concern for others. In ‘Contractualism and utilitarianism’ (Scanlon 1982), and again in What We Owe to Each Other, Scanlon identifies the tasks which an acceptable moral theory must execute. Specifically, he claims that any acceptable moral theory must explain the method of reasoning through which we arrive at judgements of right and wrong, and must also explain why there is good reason to give judgements arrived at in this way the kind of importance, and priority, that moral judgements are normally thought to have. In order to defend the priority of the moral at the level of justification he must show that friendship is grounded in morality. But the kind of morality in which it is grounded will be simply an unwelcome source of constraint unless it makes an assumption of sociability, or a desire to be at one with others. Unless, that is, it invokes the disposition to friendship, if not friendship itself. Conclusion I began by identifying two questions which are central to What We Owe to Each Other: first, the question of whether impartialist moral theory can make room for values other than the values of right and wrong, impartially understood; second, the question of how to explain the priority accorded to the values of impartial morality in cases of conflict with other values. My initial discussion of Godwin was intended simply to indicate that any sensible form of impartialism must, somehow, make room for values other than the values of right and wrong. It is on pain of absurdity that impartialists deny the value of partial relationships such as relationships of love and friendship. Moreover, by distinguishing between two kinds of question, or two levels of thinking, impartialists can successfully demonstrate that there is room for these other values. Impartialism need not, and must not, require us to be impartial in every aspect of our lives, and in every decision we take. What is more difficult, however, is to show why, in cases where reasons of impartiality conflict with the reasons provided by other values, the reasons of impartial morality should take precedence. Here I have argued, against Korsgaard, that the sources of the obligations of friendship and the sources of the obligations of morality are indeed one. And I have argued, against Scanlon, that this single source is a concern for mutual recognition, where mutual recognition


has two facets: it defends impartial morality by explaining our sense of alienation in contexts where we fail to act on impartial principles, and it also calls upon our propensity to see others non-instrumentally. However, having invoked noninstrumental concern as the ground of morality in the first place, impartial theory cannot easily eschew it in cases where friendship and morality conflict at the practical level. The fact that friendship is itself a moral concept makes the problem yet more acute. The appropriate response to these difficulties is therefore to reverse the order of explanation: rather than taking as basic that sense of mutual recognition which implies equal respect for others, impartialists should take as basic that sense of mutual recognition which implies ties of affection and concern for others. In fact, and as indicated above, many of them do this in any case. If this understanding of mutual recognition is taken as basic, then it can explain the intractable nature of the difficult cases in a way which is not merely an endorsement of self-indulgence. It can also, though, explain the motivational importance of the requirements of morality for, as we have seen, ties of affection are insufficient for friendship, which is itself a moralised notion, and one which reflects the significance we attach to impartial considerations such as fairness, equality and autonomy. Scanlon’s book raises, in acute form, a question which is central to moral philosophy, but which has, until recently, been largely neglected. In 1976 Michael Stocker noted that ‘modern ethical theories deal only with reasons, with values, with what justifies. They fail to examine motives and the motivational structures and constraints of ethical life.’ And he went on, famously, to conclude that this neglect of motivation had generated a ‘schizophrenia’ in modern ethical theory (Stocker 1976:453–4). If, philosophically, we are now on the road to recovery, that is due in no small part to Scanlon’s What We Owe to Each Other. ACKNOWLEDGEMENTS Earlier versions of this essay were delivered at the Northern Political Thought conference, Stirling, at the Universities of Bristol, Sheffield, Melbourne, and New South Wales, at the London School of Economics, and at the ANU, Canberra. I am grateful to all the participants for their helpful and constructive comments. REFERENCES Baron, M. 1991. ‘Impartiality and friendship’. Ethics, 101/4, 836–57. Barry, B. 1995. Justice as Impartiality. Oxford: Oxford University Press. Becker, L. 1991. ‘Impartiality and Ethical Theory’. Ethics, 101/4, 698–700. Gauthier, D. 1986. Morals by Agreement. Oxford: Oxford University Press. Godwin, W. 1976. Enquiry Concerning Political Justice. Harmondsworth: Penguin Ibsen, H. 1965. Plays. Harmondsworth: Penguin


Kelly, P. 1998. Impartialtiy, Neutrality and justice. Edinburgh: Edinburgh University Press. Korsgaard, C. 1996. Sources of Normativity. Cambridge: Cambridge University Press. Locke, D. 1980. A Fantasy of Reason: The Life and Thought of William Godwin. London: Routledge and Kegan Paul. Rawls, J. 1971. A Theory of justice. Cambridge, MA: Harvard University Press. Scanlon, T.M. 1982. ‘Contractualism and utilitarianism’. Sen and Williams 1982:103–28. 1998. What We Owe to Each Other. Cambridge, MA: Harvard University Press. Sen, A. and B.Williams. 1982. Utilitarianism and Beyond. Cambridge: Cambridge University Press. Stocker, M. 1976. ‘The schizophrenia of modern ethical theories’. Journal of Philosophy, 73, 453–66.

4 Contractualism, Spare Wheel, Aggregation BRAD HOOKER

T.M.Scanlon is attentive to different points of view and strikingly fair-minded. He is also penetrating and innovative. All these virtues permeate his 1998 book What We Owe to Each Other. In addition, the book is hugely ambitious. The result is one of the most important books on moral philosophy written in the twentieth century. Each of the book’s eight chapters and the appendix contains new ideas and arguments, many of which I believe will rightly be accepted and all of which deserve careful study. But the centrepiece of the book is Scanlon’s contractualist theory of wrongness. My essay will focus on two objections to Scanlon’s contractualism. One is that Scanlon’s contractualism functions only as a spare wheel. The other concerns difficulties that his contractualism has with some cases where a choice must be made between doing what will result in a greater aggregate benefit and doing what will produce a greater benefit per person. Ambitions and Scope of Scanlon’s Contractualism Scanlon has three main ambitions for his contractualism. First, the contractualist principle should identify whatever basic unity there is in moral requirements. Second, this principle must have implications about what is morally required and what isn’t that do not severely conflict with our confident moral convictions about specific kinds of case (Scanlon 1998:70–71, 242, 382 n.61).1 Third, the contractualist principle should articulate the rational force had by the moral requirements that the principle claims to unify.2 Scanlon’s contractualist principle is: An act is wrong if its performance under the circumstances would be disallowed by any set of principles for the general regulation of behaviour that no one could reasonably reject as a basis for informed, unforced general agreement. (1998:153) This is the same set of words as in Scanlon’s hugely influential 1982 article (Scanlon 1982:110; cf. 116, 122–3; Nagel 1991: chs 4, 5, 7, 8). But, in order to accommodate recalcitrant intuitions, Scanlon’s contractualism underwent considerable change between his 1982 article and his 1998 book.

CONTRACTUALISM, SPARE WHEEL, AGGREGATION 45

In the book, Scanlon (1998:6, 171–7, 178, 270–71) holds that we should distinguish between broad morality and narrow morality. Narrow morality consists of what we rational agents owe each other. Broad morality incorporates narrow morality but extends beyond it to include such things as (a) prohibitions on some kinds of consensual behaviour that affect no other rational agents and (b) protections for animals and the environment for their own sakes. To explain broad morality, we have to add ‘that affect no other rational agents’ and ‘for their own sakes’ for the following reason. Agents doing what they want to do, either to themselves or to one another, can have a negative impact on other people. And narrow morality can forbid such behaviour where there is a negative impact on other people. Likewise, narrow morality can prohibit practices that harm animals or the environment if the harm to animals or the environment has an impact on other people. In contrast, if we think that (e.g.) acts of harming animals or the environment are morally wrong even when such acts have no direct or indirect impact on other human beings, then, in Scanlon’s terms, we are invoking broad morality. Scanlon’s 1982 article did not make this distinction between narrow and broad morality. By implication, contractualism was put forward in the 1982 article as a unifying account of all morality, not merely the ‘narrow’ part of it. The book, however, restricts the scope of contractualism. The book conscientiously acknowledges, on the one hand, that many people take morality to extend beyond the realm of what we owe one another, and, on the other hand, that contractualism cannot make sense of requirements beyond what we owe one another. Contractualism is thus restricted in the book to a thesis about narrow morality. Here we see the interplay of Scanlon’s ambitions to identify whatever basic unity there is in morality, to accord with our confident moral convictions, and to identify the rational force (or what he sometimes calls the motivational basis (e.g. Scanlon 1998:147)) of narrow morality’s requirements. In order better to accommodate certain moral convictions that did not fit comfortably with contractualism, he has now restricted contractualism’s scope by claiming that it identifies the unity only in narrow morality. And, as I shall now explain, this meshes nicely with Scanlon’s thesis about the rational force (motivational basis) of narrow morality’s requirements. Scanlon’s contractualism concerns itself with social agreement not for the sake of reciprocal advantage, but in order to express respect for the nature of rational creatures. A large part of contractualism’s appeal for Scanlon is what he calls its ‘phenomenological accuracy’. He writes: When I reflect on the reason that the wrongness of an action seems to supply not to do it, the best description of this reason I can come up with has to do with the relation to others that such acts would put me in: the sense that others could reasonably object to what I do (whether or not they actually would do so). (Scanlon 1998:155)


And a little later: The contractualist ideal of acting in accord with principles that others (similarly motivated) could not reasonably reject is meant to characterize the relation with others the value and appeal of which underlies our reasons to do what morality requires. This relation, much less personal than friendship, might be called the relation of mutual recognition. Standing in this relation to others is appealing in itself—worth seeking for its own sake. (Scanlon 1998:162) Scanlon’s picture of the rational force of narrow morality fits neatly with his limiting contractualism’s scope to narrow morality. Nothing about honouring a relation of mutual recognition is involved in a decision to protect the environment even where this will benefit no rational agent. So, if the environment deserves moral protection even where this will benefit no rational agent, then this part of morality lies outside the area covered by contractualism. Again, a relation of mutual recognition—at least in the full-blooded sense—is not really at stake when we deal with animals (especially the less intelligent ones). Thus, if animals deserve protection apart from the impact on human life, then this is another part of morality that lies outside the area covered by contractualism. Having shown how Scanlon’s picture of the rational force of narrow morality fits with his limiting contractualism’s scope, let me stress that Scanlon’s conception of the relation underlying the reason to do what morality requires does not have to do with actual consent or agreement (Scanlon 1998:155; see also Scanlon 1982:116). Admittedly, as Scanlon (1998:154) acknowledges, ‘it would be pleasant to live in actual harmony with others and to have them approve of the way we behave toward them, and it is unpleasant to be in conflict with those around us and to suffer their disapproval’. Furthermore, ‘there are many cases in which morality directs us to seek consensus or to secure the permission of others before acting’ (Scanlon 1998:155). But the relation that Scanlon points to as underlying the reason to do what morality requires is not the relation of being in actual agreement with others. His reason for not giving actual agreement that status is impeccable: The appeal of actual agreement cannot be the motivational basis of morality, since there are obviously cases in which acting morally requires one to resist the prevailing consensus about what is and is not justified. If, for example, the people who are the victims of one’s action are fully convinced that their interests are much less important than those of others, they may be quite happy with, and even grateful for, much less than is their due. But it does not follow from the fact that they (and others) accept your action as justified that that action is morally correct. (Scanlon 1998:154–5)


According to Scanlon, the relation underlying the reason to do what morality requires is the relation of being in a hypothetical agreement with others. This agreement is hypothetical because it is one that would be reached under certain specified—perhaps contrary to fact—conditions. These hypothetical idealised conditions are that others are fully informed, uncoerced and similarly motivated to find principles that no one could reasonably reject. So, for Scanlon, the relation of mutual recognition that supplies the motivational basis for complying with narrow morality is not one of actual agreement but one of hypothetical agreement under ideal conditions. What is so important about such hypothetical agreement? The standard answer is that such hypothetical agreement is thought to be the right conception of impartial justifiability. For example, John Rawls (1971:189–90) wrote: Some philosophers have accepted the utilitarian principle because they believed that the idea of an impartial spectator is the correct interpretation of impartiality… Now moral judgments should be impartial; but there is another way to achieve this. An impartial judgment, we can say, is one rendered in accordance with the principles which would be chosen in the original position [i.e. where individuals work out the content of the social contract]… Instead of defining impartiality from the standpoint of a sympathetic observer, we define impartiality from the standpoint of the litigants themselves.3 As Parfit (2003) argues, however, what is more basic than the idea of hypothetical agreement is the idea of ‘justifiability to each person’. Bearing that in mind, I think we can restate Scanlon’s position as follows. For Scanlon, the fundamental rational force of moral requirements comes from the importance of impartial defensibility. Scanlon interprets impartial defensibility as requiring justifiability to each person. Justifiability to each person, according to Scanlon, requires complying with principles for the general regulation of behaviour that no one can reasonably reject as the basis of informed, unforced general agreement. But now we need to ask whether this contractualist account of impartial defensibility has implications that conflict sharply with our confident moral convictions. Scanlon rightly accepts this test. He also rightly aspires to find a fundamental moral principle that unifies moral requirements (it is far too early to give up on finding such a fundamental principle). Still, as often noted, a trap lies in wait for anyone who tries to formulate a fundamental moral principle that coheres with our confident moral convictions. The trap is that the fundamental principle coheres with our moral convictions only because it presupposes them.


The Spare Wheel Objection One of the most prominent attacks on contractualism is that it falls into this trap.4 The objection, in other words, is that Scanlon’s contractualism functions as a spare wheel, a construct that spins but does not actually bear any weight or do any work. Or the objection might be put in terms of circularity: Scanlon’s contractualism doesn’t explain what makes wrong acts wrong but instead presupposes their wrongness. In particular, according to this objection, when deciding what it would be reasonable to reject, we at least sometimes must appeal to moral distinctions. This line of objection to contractualism is hardly new. It was often directed at Rawls’s contractualist account of justice (see, e.g., Kymlicka 1991:193). Rawls relies on moral assumptions in his design of ‘original position’ in which people are contracting with one another. Rawls thinks utilitarianism counterintuitive, and so he designs the original position, or rather the parties in it, so as to prevent contractualism yielding utilitarianism. Rawls stipulates that parties in the original position are not to assume that they have an equal chance of being each person in the society that is to be structured by whatever principles are chosen. This stipulation is what keeps his theory from favouring a basic structure that would produce the highest average utility. Kymlicka objects to Rawls’s stipulating that the parties in the original position are motivated in such a way that the contract generates the results Rawls thinks intuitively right. What work is reference to the social contract doing if we are using our moral intuitions to manipulate its design? At least in Rawls’s system, it was clear that the parties in the original position do not themselves appeal to moral intuitions when choosing principles. The charge against Scanlon is that we can make his contractualist principle come out with the intuitively right conclusions only if, when we try to operate his contractualist test, we fall back on moral intuitions that have not yet been validated by that test. Scanlon’s views about reasonable rejection must not presume the very thing they are being invoked to explain (namely, conclusions about right and wrong). Before going further, let me say something about what the spare wheel objection is not. The objection is not that we test the conclusions of contractualism against our considered moral convictions. Of course we should evaluate contractualism’s conclusions against our considered moral convictions. Contractualism had better not generate conclusions that sharply conflict with moral convictions in which we have more confidence than we do in contractualism itself. The spare wheel objection is instead that our convictions should not be used within contractualism so as to make contractualism’s conclusions match our convictions. Suppose the contractualist says not only that wrongness is determined by principles that no one could reasonably reject but also that those whose rejection of principles matters are those who recognise such-and-such


FIGURE 1 THE ROLE OF MORAL DISTINCTIONS IN CONTRACUALISM

moral distinctions and who have such-and-such moral priorities, etc. In this case, it is not that moral distinctions are being determined by what reasonable contractors agree to, but rather that moral distinctions are determining what counts as a reasonable contractor. Moral distinctions are serving as input to the contract, rather than its output. Figure 1 illustrates the point. Via (3) and (4), our moral convictions appropriately play important roles in our assessment of contractualism. What is not appropriate is for our moral convictions to be playing a role in the operation of A. Let me now turn to what may seem to be a legitimate form of the spare wheel objection. Scanlon (1998:391, n.21) quotes the following from Thomson: For my own part, I cannot bring myself to believe that what makes it wrong to torture babies to death for fun (for example) is that doing this ‘would be disallowed by any system of rules for the general regulation of behavior which no one could reasonably reject as a basis for informed, unforced general agreement.’ My impression is that the explanation goes in the opposite direction—that it is the patent wrongfulness of the conduct that explains why there would be general agreement to disallow it. (Thomson, 1990:30, fn.) Scanlon (1998:391, n.21) replies:


The contractualist formula that Thomson quotes is intended as an account of what it is for an act to be wrong. What makes an act wrong are the properties that would make any principle that allow it one that it would be reasonable to reject (in this case the needless suffering and death of the baby). I believe Scanlon’s response concedes too much. Before I explain why, let me explain what I think seduced him into making it. Thomson’s objection invoked a property that (very plausibly) entails moral wrongness. How could torturing babies to death for fun ever fail to be morally wrong? If all acts having that property are wrong, then there is a puzzle about why we need to explain their wrongness by reference to principles that no one could reasonably reject. But this might be a special feature of properties that entail wrongness. In other words, this might be a feature of absolute moral prohibitions. As widely noted, it is easiest to come up with a list of absolute moral prohibitions if these refer to doing certain things solely for certain purposes. For example, such principles might refer to torturing, killing or otherwise seriously harming others merely for fun. But, with respect to the wrongness of doing such things for such purposes, I think the contractualist account is actually quite compelling. Suppose contractors are trying to find a set of principles to agree upon. Now it would be circular if we suppose these contractors, before they start contractualist reasoning, already think that torturing, killing and otherwise seriously harming merely for fun are morally wrong. So suppose they don’t already think those things. Still, they would certainly insist that the contract contain principles prohibiting such things. Each contractor would insist on these prohibitions for the sake of his or her own security. Scanlon concedes to Thomson that what makes torturing the baby to death for fun wrong is the needless suffering and death of the baby. He thinks this precludes him from saying that what makes the act wrong is that it would be forbidden by principles that no one could reasonably reject. But, actually, the contractualist can and should adopt a two-level picture here. At one level, there are properties that count more or less strongly against the moral permissibility of any act that has one or more of these properties. Some examples are being an act that kills someone, being an act of torture, being an act that harms someone, being an act of stealing, and being an act of failing to come to someone’s rescue. Other properties count more or less strongly in favour of the moral rightness of any act that has one or more of these properties. Some examples are the property of being an act that would keep the agent’s promise in some morally permissible way and the property of being an act that would benefit someone. Some deontologists stop at that level, and claim there is no explanation of why some properties count morally against any act that has them, and others count morally in favour of any act that has them. But contractualism can offer another, deeper level of explanation. Contractualism can explain why certain properties count (more or less) strongly


against the moral permissibility of an act, and why other properties count (more or less) strongly in favour of the moral rightness of an act. The contractualist explanation is that principles no one can reasonably reject would tell agents to avoid acts with any of the first set of properties, and to try to do acts with as much as possible of the second set of properties. The relevance of these points is this. An implication of Thomson’s challenge is that we have to choose between either saying that an act is wrong because it is an act of torture or saying that it is wrong because it is forbidden by principles that no one can reasonably reject. But in fact the contractualist can affirm both of these things. The contractualist can also claim to be offering an explanatorily deeper and richer account than any deontologist who stops at the first level. Admittedly, some things Scanlon says do not fit the picture I have just presented. For example, he puts forward the idea that contractualism and the alternatives to it such as utilitarianism ‘are better described as rival accounts of the property of moral wrongness itself, rather than as differing accounts of the conditions under which actions have that property’. (Scanlon 1998: p.12) I believe this is mistaken. Indeed, Scanlon himself acknowledges that in order for different moral views (contractualist, utilitarian, divine command, and so on) to be actually disagreeing, they have to be talking about the same thing and making competing claims about it. One natural way to describe the situation is thus to say that they agree about what wrongness is, and are disagreeing about what gives acts this property. (Scanlon 1998:10) I certainly agree that this is a natural way of describing the debate. Furthermore, as Scanlon suggests (1998:10, 11), we can take wrongness to be the property of violating important standards of conduct and therefore opening up the agent to serious criticism. I would elaborate by tying wrongness to the appropriateness of negative reactions such as blame, resentment and guilt (Hooker 2000:72–5). Put crudely, morality is a set of norms telling rational agents how to behave and what behaviour to feel guilty about for doing or not doing. Rival moral theories offer different accounts of what ultimately gives acts the property of being forbidden by this set of norms. (For a far more penetrating and sophisticated discussion, see Parfit, 2003.) Contractualism is most naturally thought of as claiming that the ultimate basis of an act’s moral wrongness is that the act is forbidden by principles that no one can reasonably reject. And the attraction of two-level moral theories such as the kind of contractualism outlined here is that such theories can endorse principles whose implications match our confident convictions about which acts are right and which are wrong, and yet go on to provide an explanation of why these principles are justified. As I have suggested, the explanatory ambition of contractualism is one of its main attractions. Contractualism plausibly explains why there is a morally


important difference between stealing and borrowing but not between kissing a blue-eyed person and kissing a brown-eyed person. More generally, it should try to explain why all acts that are wrong are wrong. But contractualism cannot be successful in doing that unless it can, via its own resources, correctly identify which acts are wrong. Assessing Rules Individually or as Sets Scanlon officially eschews the idea that we start with the specification of totally non-moral grounds for reasonable rejection, and then work out what his contractualism entails. He writes: It is misleading to suggest that when we are assessing the ‘reasonable rejectability’ of a principle we must, or even can, set aside assumptions about other rights and entitlements altogether… A sensible contractualism, like most other plausible views, will involve a holism about moral justification: in assessing one principle we must hold many others fixed. This does not mean that these other principles are beyond question, but just that they are not being questioned at the moment. (Scanlon 1998:214; emphasis added) Why is it important whether contractualism assesses principles individually or as sets? For reasons I will come to shortly, I think it pretty obvious that contractualism should assess principles as sets. Scanlon, however, contends that contractualism should assess principles one at a time while taking a background of other moral principles as fixed. An uncharitable diagnosis of why he makes that contention would be this. Take a case where we are asking whether contractualism can agree with some moral principle that we firmly accept. If contractualism can start by helping itself to every moral principle except the one in question, then contractualism will have an easier time agreeing with this principle. This diagnosis of why Scanlon would favour assessing principles individually is uncharitable because it seems to invite the spare wheel objection. I do not, however, think the uncharitable diagnosis correctly identifies what drives Scanlon. After all, he points to considerations he thinks support his conclusion. And, as I will show, there are other arguments for his conclusion to which he might have appealed. Let me now turn to Scanlon’s first point in favour of assessing one principle while holding others fixed. This point is that some moral problems depend for their very existence on a framework of other moral principles. For example, suppose I’m trying to decide whether morality requires me to use my resources to help the needy, who have no right simply to take these resources away from me without my consent. Any principle about what I am morally required to do with my resources for the sake of helping needy others presupposes other moral


principles that allocated to me property rights in those resources (Scanlon 1998: 214). I think Scanlon goes in the wrong direction here. Rather than presupposing some rules about property rights and then trying to decide what duty of aid there is, we should think of the two principles (about property rights and about providing aid to others) together. Indeed, shouldn’t we consider the whole set of first-order moral principles together as a package? We can put the point schematically with reference to moral principles. Moral principle A might be best given that moral principles B, C, D, E and F are fixed, B best given that A, C, D, E and F are fixed, C best given that A, B, D, E and F are fixed, and so on. In other words, each principle in the set can look superior to every possible alternative to that principle, when the principle and the alternatives are judged in a way that takes the set’s other principles to be fixed. So we might think that no one principle in this set should be changed unless others are changed as well. But this hardly establishes the optimality or justifiability of the whole set. For all we have said, the package of moral principles A, B, C, D, E and F might be far inferior to a package of alternative moral principles, such as G, H, I, J, K and L. Let me move now to Scanlon’s second point in favour of assessing one principle while holding others fixed. It is that wide room for judgement and interpretation are needed in the application of principles. For example, judgement and interpretation are needed to decide what counts as a morally binding promise, when killing is forbidden by respect for human life, and what level of sacrifice is required in order to ‘alleviate someone’s dire plight’ (Scanlon 1998:199, 225, 246, 299, cf. 218, 356–60). Perhaps this point is supposed to be relevant because our conception of any one moral principle is supposed to be shaped by our conception of other moral principles. For example, our conception of a principle about not doing others harm may be shaped by our conception of a principle about respecting others’ autonomy. Still, this point hardly shows that it is impossible to assess a package of principles as a whole. All we need is some criterion for assessing the package —and the criterion need not come from within the package itself. The criterion might, for example, be the test of what no one could reasonably reject. I turn now to the third point Scanlon makes in favour for assessing one principle while presupposing others. It is that judgement is needed in deciding what counts as a good generic reason for reasonably rejecting a principle (218, 246). This is a generic escape clause, and definitely invites worry about circularity. Scanlon needs to be quite clear about what counts as good reasons for reasonably rejecting a principle, or our fear is that moral distinctions are smuggled into those reasons, when it is precisely these distinctions that contractualism needs to justify. Before leaving the question of whether possible moral principles should be assessed individually or as sets, let me indicate two considerations that I think might have tempted Scanlon into accepting the ‘one-at-a-time’ approach.


The first of these considerations depends on the following contrast. It is one thing to predict what would happen if one moral principle were changed, while leaving all the others fixed. It is far more difficult to predict accurately what would happen if lots of moral principles were changed all at once. Perhaps normally, the more changes we contemplate introducing, the harder it is to see accurately what the consequences would be. History should have taught us to be wary of the danger of unforeseen consequences, especially the unforeseen consequences of many simultaneous changes. This danger may induce us to eschew even attempting to foresee the consequences of many simultaneous changes, and instead stick to considering only one change at a time. That would be a mistake. For sometimes we can confidently predict that changing many principles would in fact be better than changing one. And examples can easily be found. Imagine a society where people accept both the principle that physical injury and theft are the business of no one outside the victim’s family, and the principle that any victim’s family should inflict sufficient harm on injurers and robbers to prevent family members being preyed upon again. These two principles should be evaluated as a pair. Furthermore, it is easy to see that simultaneously replacing both principles could be an improvement. Or imagine a society where a principle is accepted that parents should leave the bulk of their wealth to the oldest son. Suppose that established practice also reflects a moral expectation that, after the eldest son inherits, he will support his siblings and their children if he can. Here we have two principles, one about inheritance and the other about the eldest son’s responsibilities. Clearly it would be absurd to change either of these principles without changing the other. (And they should both be changed, whether in the name of autonomy or equality or utility.) Again, I of course admit that it is sometimes very difficult to predict the consequences of changes; and the larger the change, perhaps the more uncertain the prediction. Nevertheless, large changes in social attitudes sometimes do improve matters enormously—and predictably so. So, while acknowledging the difficulty of foreseeing the consequences of change, we must be willing to try to foresee them, and to make changes when we can confidently see that their consequences would on the whole be better than the status quo. I am not suggesting that Scanlon is unduly conservative. What I am trying to do is diagnose what could have inclined him to think that his contractualism should evaluate moral principles one at a time, rather than as sets. I mean the diagnosis above to be a charitable one. Here is the other consideration that I think might have tempted Scanlon into thinking that moral principles should be assessed individually (while holding other moral principles fixed) rather than as whole sets. Scanlon (1998:157) denies that the principles endorsed by contractualism will be limited to some ‘fixed list’. There is a way of reading ‘fixed’ there as ‘immutable’. Another way


of reading it is as ‘finite’. Suppose we read him as denying that the principles endorsed by contractualism will be limited to some finite number. Why might contractualism take that position? Contractualism is supposed to say what is or is not morally required in every conceivable specific case, and any new case may require a new, more finely tuned principle. Then, since possible cases are infinite in number, there could have to be an infinite number of specific principles tailored to them. But if the number of principles endorsed by contractualism is infinite, then they cannot be put through the contractualist test as a set. Suppose we ask whether anyone could reasonably reject principle P1 as a member of an infinite set of principles S. If principles should be evaluated by contractualism not individually (while holding other moral principles fixed) but as sets, then contractualism’s evaluation of P1 will turn on its evaluation of S. Suppose we ask what would be the effects of S on some particular person, e.g. you. P1 might leave you badly off. But S might contain some other principle P2 that would more than compensate you for the ill effects on you of P1. Might S contain some further principle P3 that would claw back all the benefits P2 brought you? Even if P3 wouldn’t do that, there can be no guarantee that S does not contain some other principle that would. There can be no guarantee of this because S has infinitely many principles. So I agree the contractualist test cannot be run on an infinite set of principles. If contractualism must hold that it endorses an infinite set of principles, then contractualism must evaluate principles individually, rather than as sets. And I suspect that another possible diagnosis of why Scanlon favours evaluating principles individually rather than as sets is that he thinks contractualism must endorse an infinite set of principles. But obviously, rather than accepting Scanlon’s conclusion that principles are to be evaluated individually rather than as sets, we might reject the view that contractualism must endorse an infinite set of principles. Let me take stock. I have discussed Scanlon’s stated reasons for holding that contractualism should assess one proposed moral principle at a time while presupposing the truth of other moral principles. After finding those stated reasons very inadequate, I considered two further arguments for assessing one proposed moral principle at a time while presupposing the truth of other moral principles. Although I think these further arguments are also unpersuasive, they seem better to me than the ones Scanlon explicitly cites. So I offer them as charitable diagnoses of why he thinks that contractualism should assess one proposed moral principle at a time while presupposing the truth of other moral principles. As I said, an uncharitable diagnosis would be that, if contractualism can start by helping itself to every moral principle except the one it is trying to justify, then contractualism will have an easier time justifying this principle. That uncharitable diagnosis invites the spare wheel objection.


Fairness One way out of the spare wheel objection is for Scanlon to embrace ‘welfarist contractualism’. Welfarist contractualism defines ‘the reasonableness of rejecting a principle in terms of the loss of well-being that a person would experience if that principle were accepted…as compared with the well-being that that person and others would experience under alternative principles’ (Scanlon 1998:243). But Scanlon eschews welfarist contractualism because of responsibility and fairness. He thinks that what I can reasonably reject will depend in part not just on how well off I am left relative to how well off others are left by alternative principles but also on whether my situation is or is not the result of my own doing. According to Scanlon, what I can reasonably reject will also depend in part on fairness. I cannot comment here on Scanlon’s complex views about responsibility, but I will on the book’s appeals to fairness. Virtually all of these appeals to fairness in the book (see Scanlon 1998:211, 212, 216, 243) suggest the following idea. Suppose not less than 99 per cent of the people must do X in order to sustain some public good. Suppose I propose a principle that my mother is the one and only person who does not have to contribute by doing X. Suppose exempting her from having to make contributions would make no one worse off in terms of well-being, since there would still be 99 per cent participation. Nevertheless, others can reasonably object on the grounds that the principle I propose favours my mother arbitrarily. To identify unfairness with arbitrariness makes attributions of fairness or unfairness wholly parasitic on deployment of other moral concepts. Suppose that aggregate good will be maximised if people over six feet tall are required to step into the front line in battle (where mortality is most likely) but we do not need to require people under six feet in the front line. Is this arrangement unfair? On the view of fairness in question, the answer depends on whether the fact that an arrangement will maximise aggregate good is an arbitrary fact or instead a moral reason in favour of the arrangement. On the view of fairness in question, what is fair depends on what considerations count as moral reasons. To be sure, making a distinction between people where no moral reasons support that distinction (that is, treating like cases as if they were unlike) is arbitrary, and such arbitrariness is a kind of unfairness. Likewise, refusing to apply relevant moral distinctions (treating unlike case as if they were like) is unfair. But can we not say something about fairness more illuminating than those two sentences? Furthermore, note that if we take fairness to be merely the absence of arbitrariness, there is no necessary conflict between fairness and utilitarianism. If fairness is merely a matter of applying and complying with all relevant moral reasons, even the simplest utilitarianism will endorse fairness. Such utilitarianism will of course be in favour of applying and complying with all and only relevant


moral reasons. It is just that this utilitarianism will hold that all these moral reasons consist in facts about utilities. For Rawls, in contrast, the central point about fairness was that it could conflict with utility. Rawls argued that the basic structure of society is fair if and only if the basic liberties are secured and the arrangement leaves the worst off with more primary goods than the worst off under any alternative arrangement would have. That was a quite different—indeed far more specific—account of fairness than Scanlon’s. Furthermore, Rawls appealed to the notion of a hypothetical contract to ascertain the content of fairness (that is, to ascertain what fairness requires). Virtually all the references to fairness in Scanlon’s book, on the other hand, suggest that fairness is a consideration that helps shape what people can reasonably reject, not a concept whose content is determined by what people can reasonably reject. I found this a surprising result, given Scanlon’s lineage from Rawls. Since I do not think Scanlon’s main line about fairness in the book is adequate, I would not be surprised if he deviates from it. And there is at least one place in Scanlon’s book where he may deviate in the direction of a contractualist account of fairness. Scanlon writes: One cannot take the same pleasure in one’s cooperative relations with others as members of the same firm or university, say, if one comes to believe that they are being asked to participate on terms they could reasonably reject, and the meaning of one’s own successes and accomplishments is undermined by the thought that they were attained on terms that were basically unfair. (Scanlon 1998:163–4; emphasis added) If we read the ‘and’ that I have italicised as meaning ‘therefore’, then Scanlon’s suggestion is that the terms are unfair because they are ones that others could reasonably reject. And if that is Scanlon’s suggestion, then this passage presupposes a contractualist account of fairness. But such an account is certainly not developed in Scanlon’s book. The book’s main line about fairness is instead the one that equates fairness with nonarbitrariness. Aggregation Utilitarianism weighs the good of any individual against the good of any other individual, the good of individuals against the aggregate good of groups, and the aggregate good of smaller groups against the aggregate good of larger groups. Utilitarianism does not necessarily favour what benefits the majority, but it does favour what produces the greatest aggregate benefit, and that will often be what benefits the majority. Thus, utilitarianism allows that the good of many can outweigh the good of the few. In contrast, contractualism is usually taken to insist that the claims of different individuals have to be weighed only against the claims of other individuals, not against aggregate claims of groups.


Suppose contractualism insisted that the principles that could not reasonably be rejected are whatever ones would result in the least bad worst-off position. By insisting on that, contractualism would contrast sharply with utilitarianism, since utilitarianism focuses on aggregate utility, not on the worst off position. Furthermore, contractualism would have a compelling answer to the spare wheel objection. The answer would be that, rather than being driven by an extensive set of moral distinctions that it should be justifying instead of presuming, contractualism is really guided by only the one consideration—priority for the worst off. Scanlon does not, however, take that line. On the contrary, he argues that agents are not morally required to go round making their every decision with the aim of benefiting the worst off. Were everyone to have benefiting the worst off as an overriding aim, then each would need to worry that he or she might be lied to, stolen from or even cannibalised by others for the sake of benefiting the worst off. Protection from dishonesty and physical aggression, even when such acts would benefit the worst off, are things anyone generally has excellent reason to want from a system of moral principles. So no one could reasonably reject rules providing such protection (Scanlon 1998:223). But now consider a principle telling agents always to do what is best for the worst off as long as this does not involve harming others or dishonesty. Scanlon implicitly rejects this principle too. He writes: ‘It would…be reasonable to reject a principle that required us, in every decision we make, to give no more weight to our own interests than to the similar interests of others. From an agent’s standpoint, such a principle would be intolerably intrusive’ (Scanlon 1998:224, cf. 203–4). For everyone except the worst off, a requirement to give no more weight to one’s own interests than to the similar interests of others is less demanding than a principle requiring one always to do what is best for the worst off. Thus, if a requirement to give no more weight to one’s own interests than to the similar interests of others would be reasonably rejectable because intolerably intrusive, a fortiori a principle requiring one always to do what is best for the worst off would be reasonably rejectable because intolerably intrusive. Nevertheless, for cases where prohibitions on physical aggression and dishonesty do not come into play, Scanlon argues for a ‘Rescue Principle’. This principle requires you to prevent terrible things from happening to others as long as this does not require more than moderate aggregate self-sacrifice. (Aggregate self-sacrifice refers to total sacrifice over the whole of one’s life, not merely in the case at hand.5) Scanlon then adds a ‘Principle of Helpfulness’. This principle requires you to help others with their important projects whenever you can do so at no significant cost to you (Scanlon 1998:224). Elizabeth Ashford argues that Scanlon’s contractualism will in fact require a much more demanding principle about aiding the very needy (Ashford 2003; see also Kamm 2002:338–9). After all, the very needy stand to lose more from the rejection of a very demanding principle of aid than those who are able to help the very needy stand to lose from accepting a very demanding principle of aid.


Maybe Scanlon’s theory can be defended against Ashford’s objection. I think such a defence would emphasise the systematic effects of principles, rather than focusing on their effects in the particular case. However, rather than taking up the issue of how demanding Scanlon’s contractualism really is, I want to focus on cases where the cost to the agent is too small to be relevant. Suppose you must choose between rescuing a larger number of people, or a smaller number, or none at all. And suppose the net cost to you of doing any one of these three alternatives is exactly the same as of doing any other. What does Scanlon’s theory imply about cases where the net cost to the agent involved in choosing each alternative is the same? For Scanlon, the normal ground someone might have for rejecting a principle is that the principle would leave him or her badly off or treat him or her arbitrarily. We have already considered Scanlon’s views on fairness. So let us set them aside here. Focus on the ‘personal’ reason each person has to reject any principle that leaves him or her badly off. This personal reason has to be weighed against every other individual’s personal reason to reject any alternative principle that leaves him or her badly off. Scanlon’s basic position is to eschew aggregation across individuals. According to contractualism: The justifiability of a moral principle depends only on various individuals’ reasons for objecting to that principle and alternatives to it. This feature is central to the guiding idea of contractualism and is also what enables it to provide a clear alternative to utilitarianism and other forms of consequentialism. (Scanlon 1998:229) When assessing principles, we are not supposed to consider the goods of many people added together. Instead, we are to compare the reasons one individual has for objecting to a principle against the reasons any other individual has for objecting to other principles. Suppose that the two islands have on them people all of whom will certainly drown unless you save them. None of these people is your friend, relative, partner, promisee, or has any other special connection to you. Does Scanlon’s insistence on pair-wise comparisons of individuals’ reasons for rejecting candidate principles mean that, given the choice between saving the many individuals on Island A from death or the sole individual on Island B from death, you must not let the combined deaths of Island A’s many individuals outweigh the death of Island B’s one individual? Scanlon acknowledges that, intuitively, it would be wrong in such a case to save the smaller number of people from death. He also accepts that this is prima facie a serious problem for contractualism (Scanlon 1998:230). However, Scanlon mounts an argument that, even where the choice is between saving one and saving two, contractualism can indeed hold that morality requires


the agent in such a case to save the greater number. Here is Scanlon’s reasoning for this conclusion (Scanlon 1998:232). Imagine that there is only one person on Island A and that there is only one person on Island B. Each will be drowned unless saved. Here, the reason that the one person on Island A has to reject any principle that does not save Island A balances out evenly with the reason that the one person on Island B has to reject any principle that does not save Island B. Now think about the case in which there are two people on Island A and only one on Island B. The only person on Island B of course has a good personal reason to reject any principle that does not save Island B. But each of the two people on Island A has an equally good reason to reject any principle that does not save Island A. Consider the reason that the first one of the two people on Island A has. This reason exactly balances out the reason that the only person on Island B has to reject any principle that does not save Island B. And the second person on Island A has a reason that should be considered. Indeed, the reason that the second person on Island A has breaks the tie between the reason of the first person on Island A and the reason of the only person on Island B. Not to accept a principle allowing the two to be saved instead of the one would be to ignore the balance-tipping claims of the second person on Island A. So far, it is crucial to the story about the two islands that, if we are allowed to save Island A rather than Island B, then Island A must have on it as many people with at least as much per person at stake in our decision as there are people on Island B with that much per person at stake. This condition is satisfied in the example where Island A has on it two people each of whose life is at stake and Island B has on it one person whose life is at stake. But Scanlon (1998:240) does not insist that the balance-tipping claims of the extra person be as strong as those of the people off-setting one another. Suppose we face a choice between either, on the one hand, saving one stranger’s life or, on the other hand, saving not only some other stranger’s life but also a third stranger from permanent blindness. Suppose there are no other morally relevant facts. In this case, certainly we should save the second stranger from death plus the third stranger from blindness rather than merely the first stranger from death. The second stranger’s life offsets the first stranger’s, and then the third stranger’s sight serves as the tie-breaker. And in fact we can generalise from that case to any case in which other things are equal and an agent could either save the lives of some number of strangers or save the lives of the same number of other strangers plus the sight of an additional stranger. So Scanlon grants that the needs of an extra person on one side can at least sometimes break what would otherwise be a tie. There has been much discussion about whether Scanlon’s granting this is a capitulation to aggregation or is otherwise somehow inconsistent with his overall approach (Reibetanz 1998; Otsuka 2000; Kumar 2001; Hirose 2001). However, I will take up a different objection.


As Scanlon admits, intuitively, just as one is required to save two people from each suffering some harm rather than one other person from the very same harm, there is some number such that one is required to save that number of people from each suffering the harm rather than one other person from suffering a slightly worse harm. Scanlon’s reply is that the distinctions on which the principles I have argued for rely are distinctions between broad categories of moral seriousness. Slight differences in what happens, such as a pain’s lasting a little longer or a person’s losing two fingers rather than three, do not make the difference between a very serious loss and a moderate one. (Scanlon 1998:238–9) In effect, Scanlon suggests that, where one harm is slightly worse than each of two others, we treat the three as all being within the same broad category of moral seriousness. This enables us to take one of the less bad harms to offset the worse harm. We then let the second of the less bad harms tip the balance. Thus, we take one instance of the slightly worse harm to be outweighed by two instances of the slightly less bad harm. As Scanlon (1998:239) acknowledges, this is a dubious move given his insistence on pair-wise comparison of individuals’ personal reasons. Other things being equal, if our choice is between saving one person from the slightly worse harm or someone else from the slightly less bad harm, we should save the one from the slightly worse harm. When we weigh the claims of different individuals against one another in pair-wise fashion, we judge the one harm worse than the other. So how can the contractualist in consistency go on to say that the it would be wrong to save the one from a slightly worse harm than the many from a slightly less bad harm? Scanlon’s fairly tentative answer draws on Kamm (1993: chs 8–10) about the moral relevance of numbers. Scanlon (1998:240) puts the argument in this way. Return to the cases where other things are equal and an agent could either save the lives of some number of strangers or save the lives the same number of other strangers plus the sight of an additional stranger. The lives on one side are offset by the lives on the other. Then the fact that saving the second set of strangers would also save an additional person from permanent blindness tips the balance. Scanlon proposes to extend that line of reasoning to the case where the choice is between saving some people from a more serious harm and saving a greater number of other people from less serious harms. Scanlon writes: It might be claimed that if the less serious harms are nonetheless morally relevant to the more serious ones this means that the principle requiring (or perhaps even permitting) one always to prevent the more serious harms in such a case could reasonably be rejected from the point of view of someone in the other group on the ground that it did not give proper consideration to his admittedly less serious, but still morally relevant, loss.


One might then argue that such an individual’s claim to have his or her harm taken into account can be met only by a principle that is sensitive to the numbers of people involved on each side. (Scanlon 1998:240) In effect, Scanlon’s appeal to moral relevance is meant to allow greater numbers to matter in some cases but not in others. There are serious problems with Scanlon’s appeal to moral relevance here. Norcross (2002:307–309) shows that Scanlon’s appeal to moral relevance can prevent full-scale aggregation only by denying the transitivity of ‘ought to be done rather than’. Partly for the same reasons, Kamm (2002:347–53) argues that Scanlon’s remarks about the relevance of numbers of harms need to be refined. But, unlike Norcross, Kamm (2002:353) proposes that whether we should pick one alternative act rather than an alternative to it can depend on what other alternatives are available. I am not confident that Kamm’s proposal is intuitively plausible; but, even if it is, she does nothing to show that her principles about moral relevance are ones that contractualism generates. If contractualism needs instead to presuppose various theses about the moral relevance of numbers of harms, then again contractualism’s ambitions are turning out to be more modest than we might have hoped. An alternative proposal made by Parfit (2003) is that contractualism should undergo deep structural changes in order to reach plausible conclusions about aggregation cases. One such change might be that priority should be given to the worst off individual (as opposed to the individual with the most at stake). The second change might be the dropping of the individualist restriction, and hence the aggregate claims of groups will count. The third change might be that we give some weight to equality of distribution. I accept that, with these changes, contractualism might agree far more regularly with our convictions about aggregation cases. But in accepting these changes, contractualism would be making very considerable moral assumptions, and to that extent scaling back on its explanatory ambitions. In addition, if contractualism accepts these changes, the contrast between contractualism and a distribution-sensitive ruleconsequentialism would fade into only a quite subtle one. Let me sum up my claims about contractualism in aggregation cases. Either contractualism can reach the right conclusions about aggregation cases or it cannot. If contractualism cannot reach the right conclusions, then it is in trouble. If it can reach the right conclusions about those cases, how many moral assumptions does it have to make to do so? If the answer is that contractualism has to make a lot of moral assumptions, then the theory’s explanatory ambitions are shrinking. That is not yet a fatal objection to contractualism—unless there is some other theory that achieves at least as much but on the basis of fewer assumptions (that possess roughly the same plausibility).


ACKNOWLEDGEMENTS Derek Parfit and Philip Stratton-Lake have repeatedly saved me from many misunderstandings of Scanlon. For other helpful comments, I am grateful to John Broome, Andrew Chitty, Roger Crisp, Jonathan Dancy, Leonard Kahn, Saladin Meckled-Garcia, Michael Morris (Sussex), Véronique Munoz-Dardé, Mike Otsuka, Julian Savulescu, John Skorupski, Jussi Suikkanen, Peter Vallentyne and Ralph Wedgwood NOTES 1. I too defend this methodology in Hooker 2000: section 1.3, and in Hooker 2002: 162–5. 2. Scanlon writes: ‘While one aim of my contractualist account is to give a general criterion of wrongness that explains and links these more specific wrong-making properties [such as being an intentional killing or the breaking of a promise], this is not its only, or even its chief, aim. It also aims to characterize wrongness in a way that makes clear what reasons wrongness provides, and this aim goes beyond saying “what makes acts wrong”, at least on the most natural reading of these words.’ (Scanlon 1998:11) 3. See more generally, Rawls 1971, sections 5, 30. See also Barry 1995. I discuss the place of impartial justifiability and the hypothetical agreement conception of it in Hooker 2000:23–9, 66–70, 101–4. Incidentally, I think Scanlon makes a revealing slip when he writes: ‘Quite impartial reasoning about the rejectability of principles leads to the conclusion that we are not required to be impartial in each actual decision we make.’ (Scanlon 1998:225) From the context in which he writes this, it is clear that he equivocates on ‘impartial’. By ‘impartial reasoning’ here he means contractualist reasoning; but the second reference to impartiality in the sentence refers to strictly utilitarian impartiality. This is a testament both to the importance of impartial justification and to the naturalness of understanding impartiality in the utilitarian way. 4. For the attacks, Thomson 1990:30, n.19; Pettit 1993:302; Blackburn 1999; McGinn 1999; Pettit 1999; Kamm 2002:339–40, 342, 352; Wallace 2002:464–8; Raz 2003. For defenses of Scanlon, see Ridge 2001, Stratton-Lake 2003, and Parfit 2003. 5. For an earlier discussion of this idea, see Cullity 1995:293–95.

REFERENCES Ashford, E. 2003. ‘The demandingness of Scanlon’s contractualism’. Ethics, 113/2, 273–302. Barry, B. 1995. Justice as Impartiality. Oxford: Oxford University Press. Blackburn, S. 1999. ‘Am I right?’, New York Times, 21 February. Cullity, G. 1995. ‘Moral character and the iteration problem’. Utilitas, 7/2, 289–99. Hirose, I. 2001. ‘Saving the greater number without combining claims’. Analysis, 61/4, 341–2.


Hooker, B. 2000. Ideal Code, Real World: A Rule-consequentialist Theory of Morality. Oxford: Oxford University Press.2002. ‘Intuitions and moral theorizing’. StrattonLake 2002:161–83. Kamm, F. 1993. Morality, Mortality. Vol.1. New York: Oxford University Press.2002. ‘Owing, justifying, rejecting: Thomas Scanlon’s What We Owe to Each Other’. Mind, 111/142, 323–54. Kumar, R. 2001. ‘Contractualism on saving the many’. Analysis, 61/2, 165–71. Kymlicka, W. 1991. ‘The social contract tradition’. Singer 1991:186–96. McGinn, C. 1999. ‘Reasons and unreasons’. The New Republic, 24 May, 34–8. Nagel, T. 1991. Equality and Partiality. New York, Oxford University Press. Norcross, A. 2002. ‘Contractualism and aggregation’. Social Theory and Practice, 28/2, 303–14. Otsuka, M. 2001. ‘Scanlon and the claims of the many versus the one’. Analysis, 60, 288–90. Parfit, D. 2003. ‘Justifiability to each person?’. Ratio 16/4 (forthcoming). Pettit, P. 1993. The Common Mind. New York: Oxford University Press.1999. ‘Doing unto others’. Times Literary Supplement, 25 June, 7–8. Rawls, J. 1971. A Theory of justice. Cambridge, MA: Harvard University Press. Raz, J. 2003. ‘Numbers, with and without contractualism’. Ratio 16/4 (forthcoming). Reibetanz, S. 1998. ‘Contractualism and aggregation’. Ethics, 108/2, 296–311. Ridge, M. 2001. ‘Saving Scanlon: contractualism and agent-relativity’. Journal of Political Philosophy, 9/4, 472–81 Scanlon, T. 1982. ‘Contractualism and utilitarianism’. Sen and Williams 1982: 103–28. 1998. What We Owe to Each Other. Cambridge, MA: Harvard University Press. Sen, A. & B.Williams, eds. 1982. Utilitarianism and Beyond. Cambridge: Cambridge University Press. Singer, P., ed. 1991. A Companion to Ethics. Oxford: Blackwell Publishers. Stratton-Lake, P. 2002. Ethical Intuitionism: Re-evaluations. Oxford: Oxford University Press.2003. ‘Scanlon’s contractualism and the redundancy objection’. Analysis, 63/1, 70–76. Thomson, J. 1990. The Realm of Rights. Cambridge, MA: Harvard University Press. Wallace, J. 2002. ‘Scanlon’s contractualism’. Ethics, 112/3, 429–70.

5 Responsibility and Choice MATT MATRAVERS

Preamble Issues of responsibility have recently taken centre stage in political philosophy as a result of the commitment of liberal egalitarians to the idea that a just society will be one governed by two principles: a principle of compensation that ‘says that the institutions of a society should operate in such a way as to counteract the effects of good or bad fortune’, and a principle of personal responsibility that says ‘that social arrangements should be such that people finish up with the outcomes of their voluntary acts’ (Barry 1991a:142). That is, the goal of the liberal egalitarian is to seek a distribution that is chance insensitive and choice sensitive (to adapt Dworkin’s formula). And this goal puts the question of how to discriminate between choice and chance at the heart of political philosophy. This ‘luck egalitarian’ project, which encompasses discussions of the proper metric of justice (welfare, resources, access to advantages, etc.) has come under fire recently for having diverted the attention of egalitarian philosophers away from the proper concern of alleviating oppression (see Anderson 1999; for a response see Dworkin 2002) but whatever the truth of that, responsibility is a central concept in any account of social justice linked to equality of opportunity. That remains true even if it turns out that people are not responsible for very much and that ensuring fair equality of opportunity requires something close to equality of outcome. However, the centrality of choice and chance in contemporary liberal egalitarianism is problematic. As I have put it elsewhere, the distinction between choice and chance cuts in ways both too deep and too wide.1 ‘Wide’ because chance affects many areas of our lives that seemingly have nothing to do with justice. ‘Deep’ because, pushed far enough, chance seems all pervasive. What seem to be choices are often, perhaps always, dictated by a combination of features of the person (his character or personality) and features of the world (the available options, the choices of other people, physical facts and so on) that are not themselves the outcome of the person’s choice. I intend to put ‘width’ problems to one side. Perhaps they can be adequately dealt with by an account of the ‘social division of responsibility’ (Rawls 1982:


168–70; on which see Scanlon 1988:197–201). Such an account will, for example, allow that my having a big nose is not a matter of choice, but that compensation for the problems that it causes is hardly a matter for public action. Depth problems are much more serious because they potentially undermine the usefulness of the choice/chance distinction. They arise in familiar examples. Take two famous cases: first, in Dworkin’s example Louis has expensive tastes which mean that he would need extra resources in order to get the same level of welfare as others.2 He likes fine wines and plovers’ eggs. For Dworkin, Louis shows the bankruptcy of welfarist measures of equality. Louis ought to be regarded as responsible for his tastes and so not entitled to extra resources. Criticising Dworkin, Jerry Cohen argues that being true to the choice/chance distinction means that any such judgement must depend on whether Louis chose to develop his expensive tastes (Cohen 1989). Thus, for both Dworkin and Cohen the choice/chance distinction is crucial, but they differ in how they deploy it. Second, in an example made famous by Gary Watson, we learn of the life history of Robert Harris.3 Harris followed two teenagers from McDonalds where they had just bought burgers. Having caught up with them he killed them both. Afterwards, he ate their burgers and laughed about what he had done. To his accomplice he mused on the fun to be had from impersonating a police officer and going to inform the parents that their children had been murdered. Harris, it turned out, had been brought up in appalling circumstances and had suffered a level of abuse regarded as extreme even by his eight siblings, all of whom had ‘monstrous childhoods’. A child of alcoholic and abusive parents, Harris was systematically abused by both mother and father. His sisters describe how he went from a sweet, good natured child to a vicious psychopath.4 In their view, this change resulted from his treatment at home and in a federal youth detention centre where, aged 14, he had been raped twice and twice attempted suicide while serving time for car theft. The examples of Louis and Harris display one of the troubling aspects of responsibility. In both cases, the agent’s behaviour is in accordance with the agent’s character. Who is Louis? He is someone with expensive tastes (I shall describe him as a ‘snob’. Of course, having expensive tastes is not the same as being a snob so let me simply stipulate the Louis is a snob). Harris is a cruel and callous person who fails to understand why others’ suffering as a consequence of his actions is a reason not to perform those actions. To put the issue in its most simple form: these examples are an attempt to draw out two different thoughts. First, that Louis is a snob and Harris a despicable person. They are rightly held responsible for their tastes and actions. Second, it matters how Louis and Harris came to be the way that they are. Moreover, if the stories of their backgrounds are of a particular kind then our conviction that they can fairly be held responsible will be undermined.

RESPONSIBILITY AND CHOICE 67

Responsibility: Metaphysical and Moral Of course, questions of whether, and if so in what way, Louis and Harris are responsible can only be addressed given an understanding of responsibility and we might, and do, think of responsibility in two ways. One way reflects the belief that when it comes to responsibility there is some fact of the matter and it is a fact about the agent. Borrowing from Christopher Kutz: The fundamental idea…is that responsibility is a moral property of agents that consists in or supervenes upon underlying facts of agency and upon agents’ connections to the world. Such facts uniquely determine the moral desert of the agent; it is then a primary job of our moral and legal institutions to mete out to agents the response they deserve… Responsibility is a moral fact, pertaining to a relation between an agent and an object of assessment. (Kutz 2002:550) One version of this approach holds that the crucial fact on which the property of responsibility rests is that the agent genuinely chose his actions (and that choice is connected in the right way to his performing those actions). Louis and Harris then are responsible if, and only if, their actions rest on genuine choice. Our uncertainty in judging them responsible is simply uncertainty about whether we can track their actions back through such choices. This uncertainty may stem from a general uncertainty about whether they, and we, have the kind of freedom necessary to make such choices. If so, this is the traditional free will problem. The concern may, of course, be more narrow. We may not doubt that we are capable of genuine choice, but we may not think the same of Louis and Harris because of peculiarities in, for example, their upbringings. This approach captures some important aspects of our thinking about responsibility. When we make judgements of responsibility we worry about getting them right; about whether the person we are judging is really responsible or not. The details of Harris’s upbringing make a difference to us if only in making us pause before we come to judgement. The luck egalitarian focus on choice and chance is not accidental. It speaks to something very deep within us, in particular with respect to fairness. However, there is also something unsatisfactory about this approach. It is, as Kutz puts it, ‘radically incomplete’ (Kutz 2002:550). Holding people responsible, in the sense in which that matters to political philosophers, is a moral and ethical matter; a part of ‘what we owe to each other’ (as Scanlon puts it).5 It is not, or not only, a matter of some fact about the agent. That is, we think that we can distinguish the metaphysical question, ‘Does Louis possess the property of being responsible?’, and the ethical question, ‘Are we legitimately entitled to hold Louis responsible?’. These two questions are interconnected, but they are not the same. And, it is the separation of these questions that promises to rescue liberal


egalitarians from two problems that arise from their commitment to the choice/ chance distinction. The first, as we have seen, is that distinguishing choice and chance is difficult. The second, implicit in the discussion above, is that, as Cohen puts it: distinguishing ‘between responsibility and bad luck’ (Cohen 1989:922) requires an account of ‘genuine choice’ and this threatens to ‘land political philosophy in the morass of the free will problem’. Indeed, Cohen thinks, ‘it subordinates political philosophy to metaphysical questions’. That, he adds, ‘is just tough luck’, but ‘it is not a reason for not following the argument where it goes’ (all quotations from Cohen 1989:934, emphasis in the original). However, for many liberal egalitarians the prospect of political philosophy being subordinated to metaphysics is a very good reason not to follow the argument. For the goal of much recent liberal political philosophy has been to generate principles of justice without (particular controversial) metaphysical commitments. The title of Rawls’s essay ‘Justice as fairness: political not metaphysical’ captures the spirit (Rawls 1985). The reason for this, of course, is that liberal political philosophy is dominated by those with a concern for the reasonableness of pluralism; a concern that they translate into the demand that public principles be neutral between competing reasonable conceptions of the good. This is achieved, it is thought, by avoiding, or remaining agnostic about, controversial metaphysical questions concerning such things as value, personal identity, freedom, and so on. There is, then, a tension in contemporary liberal political philosophy. The aspiration is to theorise a society that is fair in two ways: first its principles of justice are such that they can be defended without reference to controversial and contested metaphysical arguments (that is, fair in relation to the fact of reasonable pluralism). Second, its governing principles of justice are a principle of responsibility (governing the outcome of choice) and a principle of compensation (governing the outcome of chance) (that is, fair in relation to choice and chance). Yet the latter kind of fairness presupposes an account of responsibility, which is to say that it presupposes an answer to a (particularly tricky) metaphysical question. Scanlon on Responsibility It is here that Scanlon’s account of responsibility might be thought to come to the rescue of the liberal egalitarian. For Scanlon promises an account that attends separately to both notions of responsibility described above; shows their interconnection; and remains metaphysically minimal. Scanlon distinguishes two senses of responsibility: attributive and substantive. To make an attributive judgement is to make a judgement as to whether the person is responsible for a given action in the sense that ‘it is appropriate to take [that action] as a basis of moral appraisal of the person’ (Scanlon 1998:148). To make a substantive judgement about responsibility is to make a judgement about what people are required (or not required) to do. More generally, judgements of


substantive responsibility are judgements about who should pick up the costs or benefits that result from some action. Attributive Responsibility For Scanlon it is ‘self-governance’ that is essential to questions of attributive responsibility. If an attributive negative judgement of an agent is warranted (it is appropriate, on the basis of some action, to take the action as the basis for a negative moral appraisal of the person) then it must be the case that ‘the process of self-governance that led to [the] action was both faulty and correctly attributable to the agent’ (Scanlon 1998:287). To be ‘faulty’—that is, morally at fault—the agent must respond to reasons that could be reasonably rejected by others or fail to respond to reasons where such a response is demanded by what we owe to each other (that is, faulty with respect to what we owe to each other). To be correctly attributable to the agent, the action must follow in the right way from the agent’s self-governance. What is required to be self-governing is, obviously, critical. Scanlon writes: ‘A person governs herself in the sense required if she is sensitive to the force of reasons and to the distinctions and relations between them and if her response to these reasons generally determines her subsequent attitudes and actions’ (Scanlon 1998:281). In addition, Scanlon requires the person’s self-governance to be stable (for the force of reasons to be part of the agent’s ‘character’). Scanlon uses this threefold test to develop an account of the usual excuses. ‘Innocent agent’ cases are cases where the ‘action is not, in the appropriate sense, attributable to the agent’ (Scanlon 1998:277). In these cases—for example where one is acting under posthypnotic suggestion, under the influence of drugs, or of sudden mental illness—what matters is that ‘the connection between the action or attitude and the agent’s judgments or character’ is severed (Scanlon 1998: 278). Innocent action cases concern the character of the action (rather than the connection between the action and the agent). ‘Ignorance and mistake of fact… coercion and duress belong to this…category’ (Scanlon 1998:279). Deficient agent cases cover those where the ‘person lacks the general capacities presupposed by moral agency’. Scanlon glosses: ‘If, as a result of mental illness or defect, a person is unable to understand and assess reasons or his judgments have no effect on his actions, then he cannot be a participant in a system of codeliberation, and must be seen, rather, as simply a force to be dealt with, like an animal’ (Scanlon 1998:280). Scanlon’s concern is to show that these categories of excuses are compatible with the truth of determinism and what he calls ‘the Causal Thesis’. He argues that the truth of determinism or the causal thesis would not mean that we lack the capacity to respond to and assess reasons, nor would it entail the existence of conditions that always disrupt the connection between the process of assessment and our subsequent actions.


So, even if one of these theses is true, it can still be correct to say that a particular action shows a person to have governed herself in a way that is morally deficient. (Scanlon 1998:281) Establishing this compatibilism with respect to attributive responsibility is an important part of Scanlon’s overall project. However, Scanlon does not believe that anything need follow from a critical attributive judgement—a judgement of moral deficiency—for the distribution of the costs of the action. Scanlon’s separation of attributive and substantive responsibility means that we can make the attributive judgement that a person is responsible free of the worry that it would not be fair for the person to be held fully responsible in the sense that she be made to pick up the full cost of the action. The following example neatly captures the point: Imagine a person who, as a result of generally horrible treatment as a child and the lack of proper early training, is both undisciplined and unreliable. If this person lies to his employers, fails to do what he has agreed to do, and never exerts himself to get a job done, he is properly criticized for these actions and attitudes. But if they render him unemployable it would not be permissible to deny him welfare support on the ground that his unemployability is due to actions for which he is responsible (Scanlon 1998:292). This example provides the basis on which we might consider Scanlon’s response to the Harris and Louis examples (and assume from this point that Louis did not choose to develop his expensive tastes). Louis does not seem to come under the normal excusing conditions. His tastes are his own and so he is, according to Scanlon, rightly an object of attributive responsibility. Is he substantively responsible? To answer that requires consideration of Scanlon’s account of substantive responsibility. Before that, what of Harris? Is he rightly subject to judgements of attributive responsibility? Assuming that Harris did not act ‘out of character’ (a crucial, although rather ill-defined element of Scanlon’s account)— and this is a reasonable assumption given that even his sisters describe him towards the end as having ‘no feeling for life, no sense of remorse’ (Watson 1987:273)—the critical question is whether he is self governing. This, as noted above, requires that he ‘is sensitive to the force of reasons and to the distinctions and relations between them’ and his ‘response to these reasons generally determines [his] subsequent attitudes and actions’. Are psychopaths ‘sensitive to the force of reasons’? As stated, yes. Although insensitive to particular considerations, psychopaths are not non-rational. Thus, Scanlon writes: Consider…a hardened criminal who commits terrible crimes. The fact that he…is someone who is unable to see the force of morally relevant reasons…


does not seem…to block moral criticism. If he commits these crimes because he does not place any value on other people’s lives or interests, what clearer grounds could one have for saying that he is a bad person and behaves wrongly. (Scanlon 1998:284, emphasis added) Even if Scanlon is right—and all the evidence is that Harris is a bad person who behaves wrongly—there is nevertheless something unsettling about the ease with which Scanlon comes to this conclusion. If someone is genuinely unable to see the force of moral reasons, surely that is reason at least to pause before thinking that person a legitimate candidate for moral criticism? Scanlon thinks not for two reasons: first, because the actions of the psychopath ‘would have implications for [his] relations with others that are at least very similar to (if not identical with) those of an agent who understood the relevant moral reasons but simply rejected them’ (Scanlon 1998:288). Second, because an inability to see the force of moral reasons is not the same as an inability to understand, for example, that a particular action will cause harm. In the latter case, the gap is between the agent’s understanding of the action and of the reasons relevant to it. In the former, ‘a person who is unable to see why the fact that his action would injure me should count against it still holds that this doesn’t count against it’ (Scanlon 1998:288: emphasis in the original). I confess, I find these arguments difficult to follow. It is true that there is a difference between, say, a dog that cannot make judgements about reasons at all and a psychopath who understands that his action will harm others, but is unable to understand the force of the reasons relevant to causing harm. The question is what difference that difference makes. Scanlon seems to use the claim that the actions of the person who us unable to understand the force of moral reasons and the actions of someone who simply rejects moral reasons are such that they have the same—or nearly the same—implications for their relations with others as grounds for thinking the difference makes no difference. However, that should surely be the conclusion not the argument. When Scanlon writes, in defence of this, that the psychopath ‘can judge’ that injuring others ‘constitutes no reason against so acting’ (Scanlon 1998:288) he begs the question. For those who think that being unable to see the force of moral reasons makes a difference think that in some sense the agent cannot judge; or, at least, cannot judge with sensitivity to the force of reasons. Similarly, in the second argument, Scanlon’s opponent can agree that the psychopath is someone who thinks that the fact that an action will injure others does not count as a reason not to perform the action. But, for Scanlon’s opponent, the critical question is why the psychopath thinks what he thinks and what the answer to this ‘why’ question tells us about his selfgovernance and his sensitivity to the force of reasons. In short, the worry is that the requirement of being ‘sensitive to the force of reasons’ is either insufficiently specified or too weak. Moreover, this worry is increased if one considers Scanlon’s view of substantive responsibility.


Substantive Responsibility Assume that Louis’s life is significantly less good than the life of others because he is unable to secure many of the only goods that provide him with pleasure. In addition, even if he did not choose his tastes, assume that Louis is rightly held attributively responsible. It does not follow from the fact that Louis is responsible in this sense that he ought not to be helped; that he alone must bear the costs of his tastes. Rather, judgements of substantive responsibility depend for Scanlon upon the opportunities available to the agent, specifically whether the opportunities to avoid the consequences were ‘adequate’ (Scanlon 1998:286). Scanlon’s account of substantive responsibility is grounded in his account of ‘the value of choice’. He uses the centrality of choice to unpack the opportunities to which individuals are entitled and the relationship of those opportunities to what we owe to each other. Briefly, Scanlon thinks that we value choice for instrumental, representative, and symbolic reasons. These three reasons for choice are displayed in the example of choosing a spouse in societies where that decision is not normally ‘arranged’. I wish to choose my wife for the instrumental reason that I think I am in a better position than anyone else to choose appropriately; for the representative reason that I want my choice to express my own ‘tastes and affections’; and for the symbolic reason that it would be ‘demeaning’ to have this choice made for me (Scanlon 1998:253). Instrumental, representative and symbolic considerations explain the importance of choice and explain why, other things equal, we have reason to reject principles that would interfere with our being free to exercise choice, but Scanlon makes the further claim that the value of choice also contributes to determinations of substantive responsibility. The precise way in which this argument is to work is difficult to determine from the text of What We Owe to Each Other. In one place Scanlon writes: ‘Where these three values [instrumental, representative, and symbolic] are significant (and there are no sufficiently strong countervailing ones), the principle that no one could reasonably reject will forbid paternalistic intervention, thus making it the person’s own (substantive) responsibility whether to risk the harms in question’ (Scanlon 1998:254; emphasis added).6 The ‘thus’ here suggests where a principle is rejected on the grounds of the value of choice the results of the choices thus protected are just because of this the substantive responsibility of the choosing agent. Yet, introducing the longest discussion of responsibility and choice in the book, Scanlon writes: ‘It at least seems that when a person could have avoided a certain result by choosing appropriately, this fact weakens her grounds for rejecting a principle that would make her bear the burden of that result’ (Scanlon 1998:256). This suggests a much weaker connection between the value of choice and attributions of substantive responsibility. The argument here runs as follows: where we have reason to protect choice and do so, and where an agent with an adequate opportunity to exercise choice does so, then the agent, other things


equal, has few grounds on which to avoid substantive responsibility for the outcomes of the choice. What is crucial, on both accounts, is that Louis and Harris must have acted in the face of ‘adequate’ choice situations if they are to be held substantively responsible. Recall Scanlon’s example of the person who is chronically unemployable ‘as a result of generally horrible treatment as a child and the lack of proper early training’. Scanlon argues that ‘he is properly criticized for these actions’, but that he is not (or, not fully) substantively responsible: ‘It would not be permissible to deny him welfare support on the ground that his unemployability is due to actions for which he is responsible’ (all quotations from Scanlon 1998:292). Scanlon’s summary of this case is that ‘he is responsible (that is to say, open to criticism) for these actions, but he cannot simply be left to bear the consequences, since he has not had adequate opportunity to avoid being subject to them’ (Scanlon 1998:292). In order to know, then, whether Louis and Harris are substantively responsible for their actions—ought, as a matter of justice, to pick up the costs for their tastes and actions—we need to know whether they had ‘adequate opportunity to avoid being subject to them’. Yet, despite the importance of this notion, Scanlon gives the reader very little indication of how ‘adequacy’ is to be assessed. The best gloss he gives is that an agent’s actions give rise to substantive responsibilities where there is a ‘fair opportunity’ to avoid them (Scanlon 1998:286), but the substitution of ‘fair’ for ‘adequate’ is not particularly helpful. That said, Scanlon provides a range of (memorable) examples and working from those it is possible to make a reasonable guess as to what he would say about Louis and Harris. In particular, both resemble the chronically unemployable man in having characters that make certain choices unrealistic—characters on the basis of which we rightly hold Louis to be a snob and Harris to be bad (that is, we make judgements of attributive responsibility)—and given that Scanlon does not think the chronically unemployable man substantively responsible it seems not unreasonable to assume that he would conclude the same about Louis and Harris. It is worth noting, although it can only be speculation, that it could be Scanlon’s reluctance to hold people substantively responsible that explains the blasé attitude to attributive responsibility apparent in his quick treatment of ‘the hardened criminal who…unable to see the force of morally relevant reasons’. If judgements of attributive responsibility are separate from judgements of who should pick up the bill then there is less reason to worry about them. Whatever the truth of that, Scanlon’s attitude to substantive responsibility is puzzling. As mentioned above, one of his main concerns is to show that attributive responsibility is compatible with both the Causal Thesis and determinism. If his account is to be complete then he needs to show that the same is true of judgements of substantive responsibility. Yet, he provides no such account. We might wonder why, if the chronically unemployable man, Louis and Harris are all not substantively responsible, because their characters are such that their choice opportunities were not adequate, anyone should be judged substantively


responsible. Why, the substantive incompatibilist might ask, is my choice set ‘fair’ or ‘adequate’ when some choices are closed to me because of my character and I had little or no control over my character? And that, of course, is the same challenge as was thrown at Dworkin and the other liberal egalitarians who would use the choice/chance distinction to underpin issues of justice (issues of substantive responsibility). Scanlon’s chapter ends with a remarkable (and lyrical) passage. He writes: We must…recognize that what separates us from [those who choose badly] is not just, as we would like to think, that we behave better and choose more wisely, but also our luck in being the kind of people who respond in these ways. In this respect our attitude toward those who suffer or are blamed should not be ‘You asked for this’ but rather, ‘There but for the grace of God go I’. (Scanlon 1998:294) Yet if that is the correct attitude, and if we assume that Scanlon does not intend by ‘the grace of God’ something that we have a fair opportunity to gain or lose, then the idea of substantive responsibility seems to have slipped from our grasp. In sum, Scanlon moves quickly in establishing attributive compatibilism perhaps because attributive judgements do not carry with them substantive judgements so there is not much harm done even if the account misses in some cases. However, all the worries and unanswered questions that arise in the attributive case are simply shifted to the substantive level and Scanlon does not provide an account of substantive compatibilism. The comments he makes about particular cases suggest that he is generally sensitive to the fairness threatening role of luck in making us people who respond in faulty or non-faulty ways to reasons. As a result, he is reluctant to hold people substantively responsible, albeit that there may be other reasons to make them pick up the costs of their tastes or actions.7 Conclusion In Volume II of his magisterial A History of the Criminal Law of England, James Fitzjames Stephen writes: My own experience certainly is, that people who commit great crimes are usually abominably wicked, and particularly murderers. I have the very worst opinion of them. I have seen something of a good many of them, and if I had not had that experience I should not have imagined that a crime which may be the result of a transient outbreak of passion indicated such abominable heartless ferocity, and such depths of falsehood as are, in my experience usually found in them. This peculiarity appears to me to be a reason, not for sparing them, but for putting them to death. If, however, when a bad man acts according to his nature, he is—as I think he ought to


be—put to death, I do not quite see why a person, who suddenly becomes bad by reason of a disease, should be in a better position than he who is bad by birth, education, and natural character. (Stephen 1883:185) It is not clear to me that we are in a better position now than were Stephen’s contemporaries to answer the challenge: why should someone bad ‘by reason of disease’ be better situated than someone bad by reason of ‘birth, education, and natural character’? Of course, the conclusion we would draw about what to do with bad people if we agreed with Stephen might not be the same as his, but that is not the point. The point is that people like Harris continue to pose a problem: they are certainly cruel, callous, and ‘abominable’. They are not insane or, in the ordinary sense, ill. Yet, the fact that they did not choose to be as they are, and yet pay such a high price, plagues us. We are, as one forensic psychiatrist puts it (Hare 1993:143), unsure whether a diagnosis of psychopathy should be the ‘kiss of death’ or the ‘kiss of life’ (and those words are well chosen given that he was discussing murderers in the USA). I began by saying that liberal egalitarians who put the choice/chance distinction at the heart of their accounts of equal treatment faced two difficulties. First, distinguishing what is down to choice and what to chance is difficult. Second, it threatens to subordinate political philosophy to metaphysics. As a result, I argued that liberal egalitarians are caught in a dilemma because of their commitment to two types of fairness: fairness with respect to chance and choice (which seems to require a metaphysical argument to underpin the distinction) and fairness with respect to the fact of reasonable pluralism (which requires eschewing controversial metaphysical arguments). My conjecture is that liberal egalitarians have warmly embraced the distinction between ‘metaphysical’ and ‘moral’ understandings of responsibility because it promises a way out of this dilemma. Thus, for Dworkin, a person is legitimately held responsible for his personality even if he could not have chosen it (Dworkin 2000:294) because holding people responsible for their personalities is just something we do. Given that this hardly seems a satisfactory account, liberal egalitarians would do well to turn to Scanlon who offers a characteristically subtle and sophisticated account of responsibility that, too, promises to take seriously choice and chance; to explain when, why, and in what ways people can be held responsible; all whilst remaining metaphysically minimal. The core of this essay aims to show that Scanlon’s account will not rescue the liberal egalitarian just as it is. Scanlon’s account of attributive responsibility is underdeveloped. To move as swiftly as he does he needs self-governance to be a weak condition, and it is. My guess is that this does not worry Scanlon because judgements of attributive responsibility do not carry consequences for substantive responsibility. Although that may ease our consciences, it hardly seems an adequate reason to set the bar for achieving attributive responsibility so low. Moreover, far from neutralising the threat of ‘metaphysical’ questions, Scanlon’s eviscerated notion of attributive compatibilism simply shifts the


questions to the substantive level. Yet, he says too little about substantive compatibilism, and what he says about particular examples may give heart to those who argue that the role of luck in our lives is such that the choice/chance distinction cuts deep. What this shows, however, is not that the mistake is in turning to Scanlon’s account of responsibility. Rather, Scanlon’s account is incomplete. For even if the chance/choice distinction is not, in the end, the single most important element in an account of distributive justice, it remains an essential component in our moral thinking and an unavoidable part of a complete theory of justice. I have argued that what is needed to fill out the account is, in attributive responsibility, a more developed sense of ‘reason responsiveness’, and, in substantive responsibility, an account of when choice situations are ‘adequate’ given the threat either that responsibility will be eroded altogether (because a person’s choice set will be limited by facts about himself or the world that he did not choose) or will be ascribed too widely (because in order to avoid the erosion of responsibility we fail to discriminate fairly between, to adapt Stephen’s categories, those who are bad by reason of disease, by reason of unchosen personality, and by reason of various kinds of mental incapacity). In doing this it seems to me unlikely that we can avoid more metaphysics than is currently fashionable amongst political philosophers. All that shows, however, is that a full account of responsibility will not, after all, provide the liberal egalitarian with the way out of the dilemma. As such an account is essential, the right conclusion is that the way out of that problem is simply to deny that the fact of reasonable pluralism requires eschewing metaphysics. ACKNOWLEDGEMENTS Earlier versions of this paper were given at the Universities of Cambridge and Stirling. I would like to thank the participants on these occasions for their comments. NOTES 1. This distinction, and the argument of this paragraph, borrow from Matravers 2002a: 34. 2. Dworkin 2000:48ff. This example, and the debate between Dworkin and Cohen described in the paragraph, is discussed at length in Matravers 2002b. 3. Watson 1987. 4. ‘Psychopath’ is a contested term and the description of Harris here as a psychopath is stipulative. For the purposes of this essay, Harris is presumed to be incapable of appreciating the force of moral and emotional claims. Psychopaths are ‘emotion blind’. They are emotion insensitive; unable to feel for real what others feel and give expression to in the use of emotional terms. Psychopaths are not (in any normal or legal sense) ‘crazy’ or ‘insane’. They do not suffer from delusions,


hallucinations, or the intense subjective distress that characterises most mental disorders. They are aware of what they are doing and why. Issues around ‘psychopathy’ have become contested because of the increasing use of ‘actuarial’ rather than diagnostic techniques for identifying psychopaths. 5. Cf. Dworkin (2000:287) who frames the question like this: ‘When and how far is it right that individuals bear the disadvantages or misfortunes of their own situations themselves, and when is it right that others relieve them from or mitigate the consequences of these disadvantages?’. 6. The passage from which the quotation comes is: ‘Reasons of all three kinds would figure, for example, in determining the shape of an acceptable principle governing paternalistic interference in a person’s life. Possible grounds for rejecting principles permitting such interference include claims that the interference they permit (a) would deprive people of the opportunity to make choices with significant instrumental value, (b) would interfere with choices that have important representative value for people as ways of shaping their lives and expressing their values, or (c) would stigmatise those who are interfered with by labeling them as immature or incompetent. Where these three values are significant (and there are no sufficiently strong countervailing ones), the principle that no one could reasonably reject will forbid paternalistic intervention, thus making it the person’s own (substantive) responsibility whether to risk the harms in question’ (Scanlon 1998: 254). 7. For example, we may have to imprison Harris in order to incapacitate him.

REFERENCES Anderson, E. 1999. ‘What is the point of equality’. Ethics, 109/2, 287–337. Barry, B. 1991a. ‘Chance, choice, and justice’. Barry 1991b:142–58.1991b. Liberty and justice: Essays in Political Theory 2. Oxford: Oxford University Press. Cohen, G. 1989. ‘On the currency of egalitarian justice’. Ethics, 99/4, 906–44. Coleman, J. & S.S.Shapiro. 2002. The Oxford Handbook of Jurisprudence & Philosophy of Law. Oxford: Oxford University Press. Dworkin, R. 2000. Sovereign Virtue: The Theory and Practice of Equality. Cambridge, MA: Harvard University Press.2002. ‘Sovereign Virtue revisited’ . Ethics, 113/1, 106–43. Hare, R. 1999. Without Conscience: The Disturbing World of the Psychopaths Among Us. New York: The Guildford Press. Kutz, C. 2002. ‘Responsibility’. Coleman & Shapiro 2002:548–87. Matravers, M. 2002a. ‘Luck, responsibility, and “the jumble of lotteries that constitutes human life”’. Imprints, 6/1, 28–43. 2002b. ‘Responsibility, luck, and the “equality of what?” debate’. Political Studies, 50/3, 558–72. McMurrin, S. 1988. The Tanner Lectures on Human Values, VIII. Salt Lake City: The University of Utah Press, Rawls, J. 1982. ‘Social unity and primary goods’. Sen & Williams 1982:159–86.1985. ‘Justice as fairness: political not metaphysical’. Philosophy and Public Affairs, 14/3, 223–51. Scanlon, T. 1988. ‘The significance of choice’. McMurrin 1988:151–216.1998. What We Owe To Each Other. Cambridge, MA: Harvard University Press.


Schoeman, F. 1987. Responsibility, Character, and the Emotions: New Essays in Moral Psychology. Cambridge: Cambridge University Press. Sen, A. & B.Williams, eds. 1992. Utilitarianism and Beyond. Cambridge: Cambridge University Press. Stephen, J.Fitzjames. 1883. A History of the Criminal Law of England: Volume II. London: Macmillan. Watson, G. 1987. ‘Responsibility and the limits of evil: variations on a Strawsonian theme’. Schoeman 1987:256–86.

6 Promises and Perlocutions MICHAEL PRATT

Introduction Theories of promise fall into two main camps. Those in the first camp conceive of promising as the exercise of a normative power, that is, a power to effect a normative requirement that one carry out a promised course of conduct.1 I will dub these ‘volitional’ theories. They conceive of promising as the communication of an intention to undertake, by that very act of communication, an obligation.2 As to how my communicating such an intention succeeds in binding me, most volitionists hold that my obligation depends on a conventional norm that renders bound those who communicate such an intention.3 Volitional accounts of promising have dominated the literature on the topic since the time of Hume. Dissenting views have been relatively few and unpersuasive. Unpersuasive, that is, until T.M.Scanlon joined the ranks of the dissenters.4 Scanlon’s is the most compelling attempt yet to explain promising as something other than the exercise of a normative power conferred by rules embedded in a social practice. Whereas volitionists regard promise-keeping as a sui generis obligation that cannot be accounted for except in terms of a distinctive social practice of keeping promises, Scanlon aligns himself with the majority of dissenters, who argue that the obligation of a promise is an unexceptional instance of a broader obligation, variously formulated, not to thwart the reliance or disappoint the expectations that we induce in others.5 Thus Scanlon argues that the wrong of breaking a promise is an instance of a large family of moral wrongs concerned with ‘what we owe to other people when we have led them to form expectations about our future conduct’ (1998:296). I dub dissenting theories of this sort ‘perlocutionary’. They hold that the immediate point of a promise is not to bind the promisor but to induce expectations, or reliance on expectations, in the promisee. True, promises generally manifest an intention to be bound, but so does letting go the crystal goblet betray a desire that it shatter. If a will to be bound is betrayed by a promise, it is because promises bind, and not vice-versa.


In this essay I argue that Scanlon’s theory fails to overcome a fundamental problem that plagues all perlocutionary accounts. The problem arises because promises inspire expectations of a special kind in their recipients, and perlocutionary theories are incapable of explaining how they do so. A promisee does not characteristically take the word of a promisor as mere evidence that she will perform the promised act because, for example, doing so is in her selfinterest.6 That is, a promisee who has been assured by a promise does not typically rely on an epistemic expectation that the promisor will, on all the evidence, actually perform her promise.7 Rather, he relies on a kind of normative expectation that the promisor will feel she owes it to the promisee to perform, even if doing so is not in her interest (cf. Downie 1985:266; Cupit 1996:101–02; Hollis 1998:11–12). Specifically, promisees rely on moral expectations that the promisor will regard it as wrong not to perform after having promised to do so. If it is correct then this description of the kind of assurance characteristically inspired by promises presents a problem for the perlocutionist, since it implies that the sort of assurance inspired by a promise cannot be generated until the promisor is bound by her word. The problem, in other words, is that in order for a promisee to be assured in the relevant normative sense, he must have some reason to believe that the promisor regards it as wrong to resile after giving her word. But on a perlocutionary account the promisor can have no such reason until she has inspired an expectation of performance (or reliance on such an expectation) in the promisee. There is a circle here, and it is vicious. One way to avoid this petitio principii is to assume that the characteristic moral expectations inspired by promises are spawned by antecedent non-moral expectations. In other words the promisee must be supplied with a reason to believe that the promisor will perform that is independent of the wrongness of the promisor failing to perform after getting the promisee to believe this. Such a reason might plausibly be provided by a social norm of promise-keeping. For example, Hume regarded it as a convention that people make promises on pain of social approbation and disrepute in the case of breach.8 If I am aware that you are aware of such a sanction, then this provides the basis for my (epistemic) expectation that you will in fact perform after promising to do so. The perlocutionist might then argue that, having induced me to have this expectation, you are now obligated under some suitably formulated perlocutionary principle to fulfil it. Being aware of this fact causes me to form the sort of normative expectation that is characteristic of promises.9 The strategy of this solution is to derive the moral expectations that are characteristic of promises from certain antecedent, non-moral expectations. But to the extent that these antecedent expectations are supposed to derive from a social practice of promise-keeping, this solution is unavailable to the perlocutionist, who insists that promises and their obligations can be explained without invoking the idea of a social norm of promise-keeping. The strategy is, moreover, flawed. Promisors need not, and typically do not, invoke any non-moral source of motivation to inspire confidence in promisees.

PROMISES AND PERLOCUTIONS 81

In promising, one reflexively appeals to another to believe that, having promised, one regards oneself as obligated to perform. As Scanlon puts it: When I say ‘I promise to help you if you help me’, the reason that I suggest to you that I will have for helping is my awareness of that fact that not to return your help would, under the circumstances, be wrong: not just forbidden by some social practice but morally wrong. (1998:306). The assurance of a promise is, in other words, moral all the way down. Scanlon proffers a subtle and elegant solution to the logical impasse that does not presuppose that promises generate antecedent non-moral obligations of any kind. I begin by setting out his perlocutionary account of promising in detail, including his proposed solution to the impasse. After showing that this solution fails, I go on to argue that his contractualist moral theory will support an alternative, non-perlocutionary principle that explains the obligation of a promise without falling prey to the circularity objection. The Principle of Fidelity Scanlon contends that the obligation of a promise is a ‘special case of a wider category of duties and obligations regarding the expectations that we lead others to form about what we intend to do’ (1998:295). These obligations are natural: they can be explained without reference to social rules or conventions. A social practice of promise-keeping might, where it exists, help facilitate the creation of the sort of expectations required to generate the obligation that arises upon the making of a promise. But even where they exist, such practices are dispensable; one could generate the obligation of a promise in a state of nature. The mainstay of Scanlon’s theory is his ‘Principle of Fidelity’. This principle (which I set out in detail later) requires that if under certain circumstances of mutual knowledge and expectation one attempts to assure another that one will adhere to a particular course of conduct, then one must adhere to that course of conduct if the assurance has been successful. Breaking promises is one way to fall foul of this principle. Scanlon derives the Principle of Fidelity from his contractualist moral premises according to which ‘an act is wrong if its performance under the circumstances would be disallowed by any set of principles for the general regulation of behaviour that no one could reasonably reject as a basis for informed, unforced general agreement’ (1998:153). The idea is that judgements of right and wrong are claims about reasons; specifically, they are claims about ‘the adequacy of reasons for accepting or rejecting principles under certain conditions’ (1998:3). To deliberate about right and wrong is to consider whether our actions ‘could be justified to others on grounds that they, if appropriately motivated, could not reasonably reject’ (1998:5).


When I ask myself what reason the fact that an action would be wrong provides me with not to do it, my answer is that such an action would be one that I could not justify to others on grounds I could expect them to accept. (1998:4) This ideal of justifiability is what gives interpersonal morality—the morality of what we owe to each other—its normative authority, or ‘reason-giving force’ (1998:3). It also gives its content. Morality motivates us to be able to justify our conduct to others on grounds that they could not reasonably reject, and right conduct is that which can be so justified to those who are so motivated. A valid principle of right action is one that could not reasonably be rejected by anyone moved by the ideal of finding general principles of conduct that could be justified to others similarly moved. Scanlon argues that the Principle of Fidelity is such a principle. Scanlon’s justification of the Principle of Fidelity reflects the general structure of contractualist moral justification, a structure imposed by the idea that interpersonal moral judgments are judgments about reasons and reasonable rejectability. Suppose we want to determine whether a principle prohibiting people from doing X in circumstances C is valid. Scanlon’s contractualism calls for us to ‘ask whether any principle that permitted one to do X in those circumstances could, for that reason, reasonably be rejected’ (1998:195). This requires that we identify the burdens that some would suffer if others were permitted to do X in C (the ‘objections to permission’), as well as the burdens that others would suffer if doing X in C were forbidden (the ‘objections to prohibition’). Then we determine whether the objections to permission are strong enough, relative to the objections to prohibition, to make it reasonable to reject any principle permitting X in C. If they are, then a principle forbidding X in C is valid. The behaviour X that the Principle of Fidelity forbids is the failure to carry out a course of conduct that one has, under certain conditions of mutual knowledge and expectation C, assured another that one will carry out. Scanlon argues that the objections to permitting X in C are sufficiently strong relative to the objections to prohibiting it to make it reasonable to reject any principle that permits it. The Principle of Fidelity is therefore a valid moral principle. What then are the objections to permitting X in C? Scanlon opts to turn this question around and to ask instead what reasons there are for wanting a principle like the Principle of Fidelity that prohibits X in C. He locates these reasons in what he calls the ‘value of assurance’. The idea of value plays a central role in Scanlon’s defence of contractualism (he devotes the entire second chapter of What We Owe to Each Other to the idea) and it is useful to pause briefly to consider his general account of value before turning to his discussion of the value of assurance in particular.


The Idea of Value Scanlon proffers an alternative to the view, widely held among philosophers, that the idea of value or goodness has a ‘teleological structure’ (1998:80). According to this popular account, states of affairs are the primary bearers of value or goodness. They have intrinsic value, that is, value that is not a function of their ‘tendency to contribute to or make possible something else which is of value’ (1998:79). Grasping the idea of value is, then, a matter of grasping which states of affairs have it and to what degree they have it. It is, moreover, natural to suppose on this view that as agents we have reason to act so as to bring about those states of affairs that have the greatest value. Being valuable or good is a non-natural property that provides us with reasons to respond to states of affairs in a particular way, namely, by promoting them. To be of value is ‘to be “to be promoted”’ (1998:87). In his rival account of value Scanlon makes two claims. First, he denies that taking something to be of value is primarily a matter of thinking that certain states of affairs are to be promoted. Take friendship, for instance. ‘We would not say that it showed how much a person valued friendship if he betrayed one friend in order to make several new ones, or in order to bring it about that other people had more friends’ (1998:88). The force of this claim is to disarm a powerful case for consequentialism. If to be valuable or good is to be something to be promoted, then it is natural to identify right actions by reference to the value of the states of affairs they tend to bring about. Scanlon’s second and, for our purposes, more important claim is that being valuable is itself not a property that provides us with reasons to respond in some favourable way to those things that possess it. Rather, to be valuable is to have other properties that constitute such reasons. That a particular novel is beautifully written is a reason to read and recommend it; that a certain student excels at your course is a reason to praise her or nominate her for an award. These properties (beauty and excellence) completely explain the reasons we have for responding in these ways to things that are of value, and it is unclear ‘what further work could be done by special reason-providing properties of goodness and value’ (1998:97). This general contractualist account of the relation of value to reasons clarifies the significance of the value of assurance to the justification of the Principle of Fidelity. The value of assurance is just the bundle of reasons that those who have been given assurances in circumstances C have for wanting those assurances, and for wanting them to be binding. I turn now to consider how Scanlon unpacks this bundle of reasons. The Value of Assurance People often want the assurances they receive to be obligatory so that they do not sustain any loss when they rely on them to their potential detriment. One


reason that I want your promise to sell me your car to bind you is that if you renege then the new tyres I purchased for the vehicle will have to be resold, at a loss. One reason that the carpenter I hire to renovate my kitchen wants my promise to pay him to bind me is to avoid expending hours of physical labour for naught. Some philosophers regard reasons of this kind as the true ground of promissory obligation, which they explain in terms of an obligation not to inflict losses on those in whom one has intentionally induced reliance (see, e.g., MacCormick 1982a; Fox & DeMarco 2001; Atiyah 1981). Scanlon, by contrast, regards the prevention of reliance-based losses as an inadequate foundation for promissory obligation. Preventing loss due to reliance is not a reason for wanting a principle that requires people to do as they have said they would. Rather, it is a reason in favour of a less onerous principle of loss prevention that requires those who induce others to expect some future performance to either perform, issue a timely warning that performance will not occur, or compensate the reliant party for his or her reliance losses (1998:300– 301). The value of assurance is comprised of reasons for wanting ‘a principle stating a duty specifically to fulfil the expectations one has created under certain conditions’ (1998:302; emphasis added). If the value of assurance does not include the value of preventing loss due to reliance, does it nevertheless include the value of facilitating reliance? People often want assurances so that they can plan, deliberate and take or forego action in reliance on them. One reason that I want the removalist to assure me today that he will move my household next Wednesday is so that I can proceed in the meantime to arrange to take that day off work. One reason that I want you to assure me today that you will sell me your house next month is so that I can proceed in the meantime to sell mine in order to pay for it. The capacity of assurance to facilitate reliance also accounts for its value in making possible non-simultaneous, mutual exchanges. Witness Hume’s wellknown example of two highly rational farmers: Your corn is ripe to-day, mine will be so tomorrow. It is profitable for us both that I shou’d labour with you today, and that you shou’d aid me tomorrow. I have no kindness for you, and know you have as little for me. I will not, therefore, take any pains upon your account; and shou’d I labour with you upon my own account, in expectation of a return, I know I shou’d be disappointed, and that I shou’d in vain depend upon your gratitude. Here then I leave you to labour alone: You treat me in the same manner. The seasons change; and both of us lose our harvests for want of mutual confidence and security. (2000:334) These losses could be avoided, and our harvests reaped, if only you were able to provide me with some assurance of your performance on which I might rely by performing first.


Scanlon does not deny that facilitating reliance is among the reasons that those who have been given assurances under circumstances C may have for wanting those assurances. But he does deny that this is an essential aspect of the value of assurance. He uses the case of the ‘Guilty Secret’ to illustrate that assurance is often desirable even when reliance is not (1998:302–303). Harold, we are told, is visiting a university for a term, where he meets up with an old acquaintance who is privy to an incident in Harold’s past that he now recalls with shame and embarrassment. Harold takes the acquaintance aside and asks him to give him his word that he will not tell anyone about the incident while Harold is in town. The acquaintance makes the promise and Harold breathes a sigh of relief. No one would doubt that Harold has reason to want this assurance to be honoured, despite that he is virtually incapable of doing anything differently, of making any plans or of taking or foregoing any action in reliance on it.10 The case of the Guilty Secret is not a compelling illustration of the contractualist moral force of the value of assurance. Most of the reasons Harold has for wanting assurance are also reasons for wanting not to be humiliated. Our sense that he has reason to want a principle that requires his colleague to honour his assurance is overwhelmed by our sense that he has reason to want a principle that forbids his colleague from gratuitously humiliating him, regardless of any such assurance. In a more recent essay Scanlon abandons Harold and the Guilty Secret in favour of a different example, apparently in order to avoid this kind of objection. In this example—let’s call it the case of the Coveted Horse—Ann promises Hal at the beginning of the racing season that she will sell him a certain horse at the end of the season, a horse that Hal wants more than any other horse.11 On these facts Hal has reason to want a principle, like the Principle of Fidelity, that underwrites Ann’s assurance that she will carry out some act even though Hal, like Harold, has no opportunity to deliberate, plan, or take or forego any action in reliance on that assurance. In contrast to Harold, however, none of Hal’s reasons for wanting such a principle also count as reasons for wanting a principle that prevents Ann from doing something, such as injuring Hal, that amounts to a wrong independently of her having promised not to do it. The case of the Coveted Horse, like the case of the Guilty Secret, is designed to lay bare the essence of the value of assurance by eliminating the possibility of reliance, while also serving to isolate the moral weight of that value by eliminating the relevance of other, non-promissory wrongs. In its light the value of assurance can be seen to have two aspects. That is, Hal can be seen to have two different kinds of reasons for wanting the Principle of Fidelity: those I will call ‘transpirational’ and those I will call ‘experiential’. The former are just Hal’s reasons for wanting it to be the case that the assured performance will transpire, and that the horse will be his. His experiential reasons, by contrast, are his reasons for wanting the experience of knowing or believing that the desired performance will occur; of resting assured that the horse will be his, and of not suffering the disappointment that would attend the thwarting of that belief.


Scanlon argues that the value of assurance is primarily transpirational (1998: 303; 2001:95). He admits that people have reason to want the peace of mind that assurances afford them, and that they have reason to want to avoid the psychological injury of disappointment, but he insists that these reasons are of only secondary importance.12 The value of assurance consists of reasons that are more basic than our reasons for wanting ‘freedom from worry, increased ability to sleep at night, and so on’ (1998:303). People in situations like that of Hal or Harold do not want the assurances they receive to be morally underwritten by a principle of fidelity just because they want to avoid anxiety and disappointment: ‘They also want certain things to happen’ (1998:303). Their most basic reasons for wanting the assurances they receive to be underwritten by such a principle are just their reasons for wanting it ‘actually to be the case’ that the assured performances will transpire (2001:95).13 Some of Scanlon’s commentators have failed to notice, or to take seriously, the basically transpirational character of the value of assurance. In their recent paper on Scanlon’s account of promising R Jay Wallace and Niko Kolodny refer to the value of assurance as ‘elusive’, inasmuch as it seems to involve both the value to the assured of believing that the assured performance will transpire, and the value to the assured of that performance actually transpiring (Kolodny & Wallace 2003:127).14 So far so good. But instead of construing these as two discrete components of the value of assurance, two classes of reasons for wanting a principle of fidelity, Kolodny and Wallace attempt to reduce the value of assurance to a singular ‘combination’ of these two. They argue that the value of assurance is ‘something like’ the singular value of knowing that the assured performance will transpire (Kolodny & Wallace 2003:128).15 Since the value of assurance is reducible to the value of this kind of foreknowledge, it is best understood by asking what it is ‘about prospective knowledge that other people will act in a certain way that makes this a condition that we have reason to want?’ (Kolodny & Wallace 2003:128). The trouble with reducing the value of assurance to the value of the prospective knowledge that assurances afford us is that the transpirational aspect of the value of assurance gets lost in the reduction. Not all of our reasons for wanting to be assured of things relate to our reasons for wanting to know that these things will eventuate.16 One of Hal’s reasons for wanting the assurance from Ann is that he wants it to transpire that she will sell him her horse at the end of the racing season. This reason does not depend on his wanting to know in advance of the end of the season that she will sell him her horse at the end of the season. Of course he values this knowledge too, for various reasons. But his reasons for wanting to know today that the horse will be his next month are distinct from his reasons for wanting the horse to be his next month. Their failure to adequately heed this distinction leads Kolodny and Wallace to overemphasise the experiential aspect of the value of assurance at the expense of the more basic and essential transpirational aspect. Thus they argue that in the


‘vast majority of cases’ the value of assurance is just its (experiential) capacity to facilitate reliance: In familiar coordination cases, for example, my assurance that you are going to do X has no value to me independently of its role as a basis for deliberation and planning: the only reason why I want to be assured that you are going to come around to my office at noon tomorrow is that we need to get together, and I need to integrate our meeting with my other activities and goals’. (Kolodny & Wallace 2003:129) Now it is of course true that there would be no transpirational value to me in your assurance that you will be at my office at noon tomorrow if your assurance were not also experientially valuable: that your assurance makes it the case that you will come by my office tomorrow at noon is probably of no value to me if your assurance affords me no basis for believing this to be so. In coordination cases like this one the transpirational value of an assurance depends on its experiential value. But this does not imply, as Kolodny and Wallace seem to suggest, that in such cases our transpirational reasons for wanting assurances are rendered unimportant, or are somehow superseded, by our experiential reasons. I want your assurance because it makes it more or less certain that you will visit my office at noon tomorrow. I value this ‘fact’ about the future just as much as I do that I can plan and deliberate on the basis of it. Kolodny and Wallace seem to want to argue that I value this ‘fact’ precisely because I can plan and deliberate on the basis of it—that my transpirational reasons for wanting your assurance are just experiential reasons for wanting to be able to plan on your noon arrival tomorrow. But if this were so then I would value a drug that causes me to believe that you will be there at noon as much as I value a principle that obligates you to be there at noon. My reasons for wanting it to be the case that you will come to my office tomorrow at noon include my reasons for wanting to meet with you, wherever and whenever, and regardless of my ability to plan that meeting in advance. Reasonable Rejection We have now unpacked the value of assurance, the set of reasons for wanting a principle that requires one to do as one has, under certain conditions C of mutual expectation and knowledge, assured another that one will do. According to Scanlon’s contractualist formula, these reasons are sufficient to justify the Principle of Fidelity so long as it could not reasonably be rejected by those who would be subject to it.17 Could the principle be reasonably rejected by those subject to it? The duty of fidelity is severe: it brooks neither a pre-emptive warning of nonperformance, nor compensation in lieu of performance. The Principle of Fidelity must therefore be sufficiently forgiving to permit one the liberty to make a wide range


of representations about one’s future conduct without moral risk. ‘Some freedom to change one’s mind is’, after all, ‘necessary for free intercourse between those who lack omniscience’ (Cohen 1933:573; cf. Finnis 1980:308). Does the principle afford this freedom? The Principle of Fidelity is built around six criteria: If (1) A voluntarily and intentionally leads B to expect that A will do P (unless B consents to A’s not doing so); (2) A knows that B wants to be assured of this; (3) A acts with the aim of providing this assurance, and has good reason to believe that he or she has done so; (4) B knows that A has the beliefs and intentions just described; (5) A intends for B to know this, and knows that B does know it; and (6) B knows that A has this knowledge and intent; then, in the absence of a special justification, A must do P unless B consents to P’s not being done (1998:304). Scanlon argues that this principle affords us adequate freedom from moral risk. It does so, he says, through the first of its six criteria, which requires that A lead B to believe that A will do P unless B consents to A not doing so.18 This criterion ensures that the duty of fidelity does not apply to bind those who indicate, expressly or otherwise, their wish to remain morally free to change their minds. Avoiding the duty of fidelity is just a matter of appropriately conditioning the expectations we create in others by, for example, punctuating our representations about future plans with riders like ‘But I may change my mind’, where this is not already clear from the context (1998:304, 403 n.6). Those who neglect to take even such minimal steps to avoid triggering the Principle of Fidelity ‘can hardly complain’ when they become subject to it (1998:305).19 On this basis Scanlon concludes that the Principle of Fidelity cannot reasonably be rejected. Scanlon’s argument that the Principle of Fidelity cannot be reasonably rejected lacks the rigour that characterises most of the rest of his book. I argue below that if the Principle of Fidelity does in fact afford the kind of moral freedom necessary to rescue it from reasonable rejection, then it falls prey to the circularity objection and cannot explain the obligation of a promise. The Logical Circle Notice that it is not necessary to make a promise in order to attract the duty of fidelity. The Principle of Fidelity is ‘not just the social institution of promising under another name’ (1998:306). The duty of fidelity obtains when there has been an inducement of assurance in the stipulated context of mutual intention and expectation, and this might be achieved in any number of ways. Thus I might convince you that I am a devoted member of the Sacred Brotherhood of Reindeer and then proceed to swear on my honour as a Reindeer that I will perform. Or I might lead you to believe that I will perform because it is in my own economic interest to do so (1998:297–8). Scanlon argues that what


distinguishes promising from these alternative means of inspiring assurance is ‘the kind of reason that the promisee has for believing that the promiser will perform’ (1998:306). What kind of reason is this? How does a promise token such as ‘I promise to do P’ assure a promisee of P so as to trigger the Principle of Fidelity and bind the promisor? In addressing this question Scanlon runs up against the same logical problem we observed earlier, the circularity that afflicts all perlocutionary theories of promise. The Principle of Fidelity will bind a promisor only once she has succeeded in assuring the promisee of P. But how can she succeed in assuring him of P without first binding herself to do P? As I argued earlier, the only way to exit this logical circle is to supply the promisee with a reason to be assured of performance that is independent of the wrongness of the promisor failing to perform after getting the promisee to be assured of this. One might, for example, allow for the possibility of spawning moral assurances from non-moral ones, such as those generated by a social practice of promise-keeping.20 Once such a social practice is taken for granted and the capacity of promises to generate nonmoral assurances is assumed, then the Principle of Fidelity may be able to explain how promises also inspire, in turn, their characteristic moral assurances. As we have seen, Scanlon does not subscribe to this kind of strategy, since he believes that the expectations generated by promises are moral from the bottom up. He insists, in other words, that it is not necessary to invoke a social practice of promising, or any other non-moral source of expectations, in order to explain how promises generate their characteristic moral assurances. ‘I do not doubt’, he remarks, ‘that there is such a thing as a social practice of promising’. But he is emphatic that this practice ‘plays no essential role’ in explaining why breaking promises is wrong (Scanlon: 296). His argument to this effect is elegant but, I will argue, ultimately mistaken.21 An Attempt to Exit the Circle Scanlon begins his argument by suggesting that the function of a promise token such as ‘I promise’ is like that of ‘trust me’. We do not utter such expressions with a view to invoking a social norm of promise-keeping. Their purpose is rather to communicate a ‘settled’ intention to carry out some act that the promisor knows the promisee desires, and also to convey the promisor’s regard for the Principle of Fidelity, for the contractualist reasoning that underlies it, and for the ‘general moral fact’ that it would be wrong not to carry out the desired act after convincing the promisee of this intention under the circumstances (1998:308– 309). Accepting this dual claim of intention and moral awareness gives the promisee reason to ascribe to the promisor the sort of moral motive to perform that is characteristic of promises. But this is a sleight of hand. The ‘general moral fact’ that Scanlon identifies as the promisor’s reason for performing her promise, is not a fact at all. It would not be wrong for the promisor to fail to adhere to her intention to perform after


convincing the promisee of that intention, and to the extent that it would make such a failure wrongful, the Principle of Fidelity is not a valid moral principle. In other words Scanlon’s attempt to escape the circularity lands him on the horns of a dilemma since it can succeed, I will argue, only if the Principle of Fidelity fails as a moral principle. Let us consider in detail Scanlon’s argument that the sort of assurance required to engage the Principle of Fidelity can be inspired by a promise. His argument is constructed around the ‘Principle of Manipulation’, which defines another in the same family of wrongs as that defined by the Principle of Fidelity. The Principle of Manipulation establishes that it is wrong to convince another that one intends to do something that one knows the other cares about and may rely on being done, if one does not actually have this intention (1998:298). It is presumably also wrong to attempt to do this, and so it follows that it is wrong to utter the words ‘I promise to P’ in such circumstances without actually having an intention to do P. By making a promise, therefore, I give you reason to believe that I have an intention to perform, and hence a reason to believe that I will perform (1998:308–309). This explains’, concludes Scanlon, ‘how the expression “I promise” can be used to create an expectation and thereby an obligation’ even in the absence of a practice of promise-keeping, or any other non-moral source of motivation (1998:309). But if this argument is correct and the Principle of Manipulation can account for the capacity of a promise to inspire the sort of expectations required to trigger the Principle of Fidelity, then the latter principle is not morally justifiable. Consider the kind of expectations that are warranted by the Principle of Manipulation. That principle will warrant a belief that the promisor has the intention to perform that she is attempting to persuade the promisee she has. It supplies no reason, however, for believing that the promisor regards herself as morally constrained by her promise to adhere to that intention. It would not, after all, be contrary to the Principle of Manipulation for me to change my mind and opt not to do P after promising to do so, as long as I intended to do P at the time I made my promise. Scanlon’s argument from the Principle of Manipulation implies, therefore, that the Principle of Fidelity can kick in to obligate a speaker to perform who has given the hearer no reason to believe that she, the speaker, believes that having said she would perform, she now owes it to the hearer to do so. And if this is the case then the Principle of Fidelity is, I will argue, not a justifiable moral principle.22 The Principle of Fidelity, if it is a valid moral principle, must require that the speaker give the hearer reason to believe that she, the speaker, believes that having said she would perform, she now owes it to the hearer to do so. To see why this is so, recall that the principle must, if it is morally sustainable, be precluded from applying to bind a speaker who communicates a desire to remain morally free to change her mind. If the principle could apply to bind such a speaker it would, as we saw, be grounds for reasonable rejection. I will argue that the Principle of Fidelity is precluded from applying to bind such a speaker


only if it requires that the speaker give the hearer reason to believe that she, the speaker, believes that having said she would perform, she now owes it to the hearer to do so. I make this argument by way of an example.23 Suppose you want a ride to work tomorrow, and you ask me if you can count on me to give you one. I reply that in all likelihood I will be able to give you a ride, and that I fully intend to do so—‘but’, I add, ‘please understand that I do not want to be taken to have bound myself to do so, and I do not regard myself as so bound’. Now if I incur by this response any moral obligation at all, it is not an obligation of fidelity.24 It is not, that is, an obligation to give you a ride to work tomorrow. If it is morally justifiable, then, the Principle of Fidelity cannot apply to bind me in this case. What would preclude it from doing so? The answer must be that the qualifying rider in my response (to the effect that I do not regard myself as obligated by my utterance) operates to preclude you from forming the kind of expectation necessary to trigger the principle. But what kind of belief could be precluded by my counselling you not to regard me as obligated to pick you up for work tomorrow? It is not a belief that I will give you a ride to work. If you had reason to believe this prior to my statement, then I have said nothing to disabuse you of that belief. And if you had no reason to believe this prior to my statement, you have at least some reason to believe it now, since you have been convinced of my present intention to do as I have said.25 The only relevant belief that is plausibly precluded by my rider, I suggest, is just your belief that I think that, having said what I said, I now owe it to you to pick you up for work tomorrow. In other words, the Principle of Fidelity can be justified only if it is understood to require that the speaker give the hearer reason to believe that she, the speaker, believes that having said what she said, she now owes it to the hearer to do as she said she would. But in the absence of some social practice of promise-keeping, a promisee would have no reason to ascribe such a belief to a promisor. It follows that if the Principle of Fidelity is a valid moral principle then Scanlon’s argument from the Principle of Manipulation cannot explain how promises can generate the kind of assurance required to trigger the Principle of Fidelity. The Conditional Belief Requirement Scanlon would object that my argument ignores the critical first criterion in the Principle of Fidelity, which requires that the speaker intentionally cause the hearer to believe that she will perform unless the hearer consents to her not performing. It is this conditional belief requirement (as I shall call it) that precludes the principle from being triggered in my example—not a requirement that the speaker cause the hearer to believe that she thinks she owes it to the hearer to perform (1998:403 n.6, 404 n.9). Now Scanlon may be correct that the conditional belief requirement precludes the Principle of Fidelity from being engaged in my example. But I will argue that


if it does have this effect, it is only because that requirement amounts to a restatement of the requirement that I am alleging to be necessary. How might a rider to the effect that I do not regard myself as bound by my utterance preclude you from believing that I will perform unless you consent to my not performing? How could your knowledge that I do not regard myself as bound by my utterance preclude you from believing that my performance is conditional on your not consenting to my nonperformance? How, in other words, could such knowledge preclude you from believing that my reasons for performing will lose their force if you consent to my nonperformance? Presumably this knowledge could preclude you from believing this only if you suppose that the only reason I could possibly have for performing that could be extinguished by your consenting to my nonperformance is that which you know I avowedly do not have—namely, that I owe it to you to perform. In order to satisfy the conditional belief requirement and trigger the Principle of Fidelity, therefore, the hearer must ascribe to the speaker a belief that she, the speaker, owes it to the hearer to perform. In other words the conditional belief requirement can only do the work Scanlon requires of it if it amounts to a restatement of the requirement that I am alleging to be necessary. This argument is borne out by reflecting on the kind of belief that is entailed by Scanlon’s conditional belief requirement. What is it for me (a hearer) to believe that you (a speaker) will perform unless I consent to you not performing? In particular, what kind of reason am I supposed to ascribe to you for performing that depends for its force on my not consenting to your nonperformance? It seems that the only plausible reason of this description that I might ascribe to you is that you owe me a promissory obligation to perform. Such an obligation will, after all, cease to obtain if I consent to your nonperformance—I can release you from your bond. It is difficult to imagine any other reason that could be similarly extinguished in this manner. A Different Attempt to Exit the Circle I have argued that Scanlon’s argument from the Principle of Manipulation that promises can inspire the kind of expectations necessary to trigger the duty of fidelity lands him in a dilemma, since if that argument is correct then the Principle of Fidelity is not a valid moral principle. That principle must, if it is valid, permit me the moral freedom to change my mind where I have, explicitly or otherwise, reserved the right to do so. If it is to afford me this liberty, moreover, it must require that the speaker induce the hearer to believe that the speaker believes that, having said what she said, she now owes it to the hearer to perform. But such a belief is not warranted by the Principle of Manipulation. So if Scanlon is right that this principle can explain how the kind of assurance required by the Principle of Fidelity can be inspired, then the latter principle must be morally unsustainable.


Scanlon’s argument from the Principle of Manipulation appears, however, to be supplemented by a different attempt to explain how promising can generate the expectations necessary to engage the Principle of Fidelity. This alternative argument, which is less fully and less explicitly developed than the first, is premised on Scanlon’s view that in promising one communicates one’s regard for the Principle of Fidelity and for the moral reasoning that underlies it (1998:306– 307). It is this information that inspires assurance in a promisee. By attempting to induce assurance while admitting that she is in a situation in which if she succeeds in doing so she will attract a duty of fidelity, the promisor thereby induces assurance, thus attracting the duty. She acknowledges that an obligation is being created, and by so doing creates one.26 This accounts for our sense that a promise involves the communication of an intention to be bound. But instead of adopting the volitional view and locating the binding force of a promise in this act of communication, which would seem to presuppose a social practice that renders those who communicate such intentions bound, Scanlon ingeniously locates it in the assurance which that communication makes possible. But the strategy fails. Consider how a promisee’s assurance is supposed to be facilitated under this strategy. The idea seems to be that the promisee has good reason to accept the promisor’s invitation to be assured since the promisor admits that she will then regard herself as bound. Accepting the invitation involves an upside with no risk. But this will not work. That a belief would prove useful to me cannot supply me with a reason for having it. As Michael Robins puts it: There may be all kinds of expectations and beliefs which, if we only had them, would make the world a far better place in which to live. But wishing we had them—no matter how rational we may be—would not in the least lead us to have them. (Robins 1976:325) Govert den Hartogh makes this point neatly. It may be that when I wake up in the morning believing that I will have a good day, my day goes well as a result. Even so, I cannot just go ahead and form that belief in order to generate this salutary effect. ‘That’, says den Hartogh, would be ‘like taking a pill for its placebo effect’ (den Hartogh 1998:369). Reconstructing the Principle of Fidelity If the criticism I have advanced is sound then it is enough to defeat Scanlon’s theory of promissory obligation. Notice, however, that my criticism leaves the contractualist reasoning on which the Principle of Fidelity is based fully intact. This is important, because when that contractualist reasoning is examined it becomes apparent that it is capable of supporting an alternative principle of fidelity that escapes the force of my criticism. In this section I sketch such a principle and show how it might be justified under Scanlon’s contractualism.


My criticism of the Principle of Fidelity fastened on its requirement that the speaker cause the hearer to be assured of some performance P. I argued that in the absence of a social practice of promise-keeping, and assuming the Principle of Fidelity is morally valid, there is no way to assure another in the pertinent sense by making a promise. It is unclear, however, why Scanlon considers it necessary to restrict the application of the principle in this way. The requirement that assurance be successfully rendered before an obligation obtains is not ordained by the contractualist reasoning that supports the principle. According to that reasoning, the Principle of Fidelity is justifiable because it secures or underwrites the value of assurance in a manner that could not reasonably be rejected by those who would be constrained by the principle. But why does the principle secure this value only for those who are already assured of the relevant performance? What contractualist moral work is done by the requirement that one actually succeed in one’s attempt to get another to expect performance, before one becomes obligated to perform? I shall argue that the value of assurance is capable of justifying an alternative, non-perlocutionary principle of fidelity that would obligate those who merely issue assurances of an appropriate kind. Before proceeding to make this case I attend to an objection raised by Kolodny and Wallace to an earlier version of my argument. They insist that a principle capable of binding a speaker to P who has not succeeded in causing the hearer to be assured of P cannot, therefore, be justified by reference to the value of assurance. As they put it, in summarily dismissing the viability of any such nonperlocutionary principle of fidelity (‘F’): ‘If F can be triggered in cases in which no assurance has been produced, then it becomes hard to maintain that the justification of F lies in the value of assurance’ (Kolodny & Wallace 2003:135). The authors offer no support for this assertion, perhaps because they regard it as obvious. It is, however, mistaken. Why would the fact that a principle of fidelity could be triggered even where no assurance has been produced preclude it from drawing valid support from the value of assurance? This would be true only if the value of assurance consisted only of our reasons for wanting to continue to be assured once assured. If this were true, then it would indeed be impermissible to invoke the value of assurance in support of a principle that binds people who have not caused others to be assured. Kolodny and Wallace appear, that is, to interpret assurance extremely narrowly, as an experience that is not of any value except insofar as it is already being enjoyed by the valuer—as an experience that cannot be valued unless one is already experiencing it. For if the experience of being assured could be valued by someone not already experiencing it, then surely a non-perlocutionary principle of fidelity designed to facilitate or bring about the experience of being assured could draw valid support from the value of assurance. Moreover, surely the experience of being assured is an experience that can be valued before one experiences it—just as someone’s friendship is something I can value before I secure it. Surely, therefore, the value we place on the


experience of being assured can support a principle of fidelity that affords us that experience in certain circumstances in which we are not already assured. Having dealt with this objection I proceed now to defend my thesis that that the value of assurance is capable of justifying an alternative, non-perlocutionary principle of fidelity that would obligate those who merely issue assurances of an appropriate kind. Recall that the behaviour X that the Principle of Fidelity proscribes is the failure to perform some act P which is the subject of an assurance given under circumstances C, where C includes the circumstance that the recipient of the assurance has been rendered assured by it. As we saw, Scanlon argues that the objections to permitting X in C are sufficiently strong, compared to the objections to prohibiting it, to make it reasonable to reject any principle permitting X in C. Now let circumstances C* be substantively identical to C except that they do not include the circumstance that the attempted assurance has been successful. That is, C* represents the following state of affairs: (1*) A issues an assurance to B that A will do P; (2*) A knows that B wants to be assured of this; (3*) A acts with the aim of providing this assurance; (4*) B knows that A has the beliefs and intentions just described; (5*) A intends for B to know this, and knows that B does know it; and (6*) B knows that A has this knowledge and intent. My claim is that the objections to permitting X* in C* are just as strong relative to the objections to prohibiting it as the objections to permitting X in C are relative to the objections to prohibiting it. In particular I want to argue that every significant objection to permitting X in C counts as an objection to permitting X* in C*, and that every significant objection to prohibiting X* in C* counts as an objection to prohibiting X in C. If I am right, then if it is reasonable to reject any principle permitting X in C, it must also be reasonable to reject any principle permitting X* in C*. Consider first whether there are any significant objections to permitting X in C that do not count equally as objections to permitting X* in C*. Consider, that is, whether people like Hal or Harold who have been assured of some performance P in circumstances C have any significant reasons for wanting a principle making P obligatory that depend on the fact that they have already come to believe that P will be performed. Have they, in short, any significant expectation-dependent reasons for wanting a principle of fidelity? Before proceeding to answer this question, it is well to emphasise that what I am calling ‘expectation-dependent’ reasons for wanting a principle of fidelity are just those I referred to above as the reasons we have for wanting to continue to be assured once assured. They do not include our reasons for wanting to become assured of performances that we are not yet assured of. For example, our reasons for wanting a principle that would facilitate reliance by underwriting the assurances we receive, giving us grounds on which to become assured by them,


and thereby enabling us to plan and act on them—these reasons are not expectation-dependent. Reasons of this kind—viz., the kind of reasons that Hume’s farmers have for wanting a principle of fidelity—bear just as powerfully against permitting X* in C* as they do against permitting X in C.27 Our question, recall, was whether there are any significant expectationdependent reasons for wanting a principle of fidelity. It may seem that there are, since someone who has been led to expect that P will be performed is liable to act to his prejudice in reliance on that expectation, giving him a reason to want P to be performed that depends on his having been convinced of that performance. But as we have seen, reasons for wanting to prevent loss or harm due to reliance are not good reasons for wanting a principle binding people to fulfil the expectations they encourage. It may be urged that even if this is so, recipients of assurances who have been convinced of performance are psychologically vulnerable to nonperformance in a way that recipients who form no such expectation are not. Making P obligatory would protect such people against the risk of suffering a kind of disappointment, providing them with an expectation-dependent reason for wanting a principle of fidelity. As we have seen, however, Scanlon does not regard the prevention of psychological disappointment as a morally significant aspect of the value of assurance.28 It is easy to see why. From the point of view of someone who has been led by an assurance to expect P, the risk of psychological harm due to dashed beliefs or expectations is typically of trivial concern. People in this situation will of course be concerned about the risk that they will be disappointed. But this is just another way of saying that they will be concerned about the possibility that they will not get what they are after, namely, P. Indeed, it is arguable that our reasons for wanting to prevent the experience of disappointment that attends the breaking of a promise are not valid reasons for wanting a principle that compels promisors to perform. They seem much more apropos a principle requiring those who disappoint expectations to apologise or otherwise try to assuage hurt feelings. I conclude that there are no significant objections to permitting X in C that do not also count as objections to permitting X* in C*. In other words there are no significant expectation-dependent reasons in favour of a principle of fidelity. Consider now the second aspect of my claim: that there are no significant objections to prohibiting X* in C* that do not also count as objections to prohibiting X in C. Consider, that is, whether a non-perlocutionary principle that prohibits X* in C* would impose any burdens on speakers that the Principle of Fidelity would not impose on them. Both principles require the same thing of their obligors, namely, strict performance. If there is a difference in the burdens they impose, then, it must relate to the relative degree of freedom from moral risk that each affords. Now we have seen that if the Principle of Fidelity is valid, it must afford those who state an intention to perform some future act while evincing a desire not to be bound


by that statement the moral freedom to change their minds. Would an alternative non-perlocutionary principle of the sort I am advocating permit speakers this kind of moral freedom? The answer depends on the kind of ‘assurance’ that is required in order to engage the principle and bind the speaker. The necessary moral freedom could be secured by defining ‘assurance’ as the volitionists define ‘promise’—that is, by requiring that a speaker communicate an intention to undertake, by that very act of communication, an obligation to do some future act P.29 Conclusion I began this essay by asserting that theories of promise fall into two main camps, volitional and perlocutionary. I conclude it by defending a theory of promise that represents a via media between them. I have shown that every significant objection to permitting X in C counts as an objection to permitting X* in C*, and that every significant objection to prohibiting X* in C* counts as an objection to prohibiting X in C. It follows that since it is reasonable to reject any principle permitting X in C, then it must also be reasonable to reject any principle permitting X* in C*. The following alternative principle of fidelity, prohibiting X* in C*, cannot, therefore, reasonably be rejected: If (1) A communicates to B an intention to undertake, by that very act of communication, an obligation to do P; (2) A knows that B wants to be assured of this; (3) A acts with the aim of providing this assurance; (4) B knows that A has the beliefs and intentions just described; (5) A intends for B to know this, and knows that B does know it; and (6) B knows that A has this knowledge and intent, then, in the absence of a special justification, A must do P unless B consents to P’s not being done. A contractualist account of promising structured around this non-perlocutionary principle escapes the logical difficulty associated with perlocutionary accounts without abandoning their insight that promissory obligations are natural obligations that do not depend on the exercise of a socially-constituted normative power.30 There is something unsatisfying about a theory that cannot explain the moral force of a promise except in terms of a conventional norm that confers a power on people to voluntarily assume obligations. The wrong I do in breaking a promise involves a form of mistreatment. It is a wrong I do to another, and it is hard to see how that sort of wrong—how a kind of mistreatment—can depend for its occurrence on a social rule that enables one to voluntarily place oneself under a duty. On a contractualist account, the mistreatment involved in breaking a promise consists in the failure to respond properly to an important human value, the value of assurance. A proper response to that value is one that would be permitted by


any principle designed to protect it—any principle of fidelity—that could not be reasonably rejected by anyone moved by the ideal of finding general principles of conduct that could be justified to others similarly moved. Scanlon posits the Principle of Fidelity as such a principle. I have argued that, to the contrary, this principle is either invalid or else it is incapable of explaining the obligation of a promise. In its stead I have defended a non-perlocutionary principle of fidelity that escapes this criticism, and which protects the value of assurance more faithfully. NOTES 1. See, on normative powers, Raz 1999:97–106. 2. Among those who adopt this or similar definitions of promising are David Hume (2000:331–7); Joseph Raz (1972, 1977, 1982); John Finnis (1980:298–308); John Rawls (1971:344–50); John Searle (1969:56–64; 2001:193–200); H.L.A.Hart (1994:43–4, 225); Charles Fried (1981:7–21); Elizabeth Anscombe (1978); and Tony Honoré (1999:59). 3. Joseph Raz is a notable exception: see Raz 1972:96–7. Volitionists differ as to how the social norm of promise-keeping generates the moral obligation of a promise. Hume, for instance, regarded fidelity to promises as a virtue because of the sense of moral approval inspired by acts of adherence to the socially-useful convention of promise-keeping. For Rawls, by contrast, the moral force of a promise derives from a principle of fairness according to which those who benefit from a just social practice such as promise-keeping are bound to adhere to its rules. These differences need not detain us. Our concern here is with what defines volitionists as a class, namely, their conception of promising as the exercise of a normative power to bind oneself to a promised course of conduct. 4. 1998:295–327, a slightly-revised version of Scanlon 1990. 5. For classic examples, see Mill 1961:584, and Sidgwick 1962: Bk III, VI, Sec.5. Recent perlocutionary accounts include MacCormick 1982; Fox & DeMarco 2001: 131–55; Atiyah 1981; Weale 1978:67–8; Postow 1980:65. 6. I will use the feminine pronoun to refer to promisors/speakers, and the masculine to refer to promisees/hearers. 7. To the extent that promisees do rely on epistemic expectations of performance, these expectations will typically depend on just the sort of moral expectations of performance that I am arguing are characteristic of promises. My epistemic expectation that you will perform your promise is inspired by my belief that you tend to behave morally, and thus that you are likely to act on your sense that it would be wrong not to perform after promising to do so. 8. ‘When a man says he promises any thing he…subjects himself to the penalty of never being trusted again in case of failure’ (Hume 2000:335). 9. For an example of an argument along these lines, see McMahon 1989. 10. Scanlon stipulates that there is nothing Harold can do to shield himself from the embarrassment that would result from divulgence of the secret. ‘He can’t leave town, and I am assuming that murder and bribery are ruled out. He’s not that kind of person’ (1998:303).


11. 1998:93. I have added the names of the parties. 12. In a recent essay Scanlon alludes to the argument that broken promises inflict a psychological sense of injury, and admits that one thing people in situations like Hal’s want is ‘a confident state of mind’, but then proceeds to argue that what such people primarily want is for certain things ‘actually to be the case’ (2001:95). 13. ‘What I am calling the value of assurance reflects reasons of [both the experiential and transpirational varieties], but it is the latter that are primary’ (2001:95). 14. ‘On the one hand, the value to Harold of the assurance that you will not reveal his secret is not simply the value to him of his believing that you will not reveal his secret… On the other hand, the value to Harold of the assurance that you will not reveal his secret is not simply the value to him of your not revealing his secret either’ (Kolodny & Wallace 2003:127). 15. Thus they write that the value to Harold of his assurance is ‘some combination of his secret’s being kept and the value to him of the belief that his secret will be kept. The value to him of assurance that you will keep his secret, one might say, is something like the value to him of knowledge that you will keep his secret’ (Kolodny & Wallace 2003:128). 16. The authors qualify their use of the word ‘knowledge,’ noting that it may not be possible strictly speaking to know that X, if X is a future state of affairs (Kolodny & Wallace 2003:128). 17. ‘The reasons potential promisees and promisers have to want such a principle of fidelity [are] in my view sufficient to establish it as a duty unless it would be reasonable for potential promisers to reject such a principle’ (1998:304). 18. The parentheses in the first criterion of the Principle of Fidelity are curious, given their importance to the principle’s justifiability. Scanlon elsewhere emphasises these parenthetical remarks: see 1998:403 n.6. 19. Scanlon is not suggesting that those who fail to take the necessary preventive measures forfeit their right to complain about the principle. Rather, they cannot complain because the Principle of Fidelity affords them ample choice not to be bound, and the ability to choose what obligations we come under is something we have good reason to want. Scanlon’s discussion of the value of choice is relevant here. See 1998:251–67. 20. Kolodny and Wallace solve the circularity problem differently, by arguing that the social practice of promise-keeping generates a kind of moral obligation—viz., an obligation not to free-ride on the practice—which in turn generates the sort of assurance necessary to trigger the Principle of Fidelity (Kolodny & Wallace 2003: 144–54). 21. It might be objected that the argument I go on to make is not what I suggest it is, namely, a refutation of Scanlon’s principal claim that the wrong of promisebreaking is independent of a practice of promise-keeping. My argument amounts to a refutation of the quite different claim that promises can generate assurances in the absence of such a practice. In other words, I unduly conflate the problem of explaining why promise-breaking is wrong (which is what Scanlon is concerned with) with the problem of how promises generate assurances. The two are very different problems. In general these are different problems. But in the particular context of Scanlon’s account of promising these problems are very closely related. Promise-breaking is wrong on his account only if promises can generate assurances. So promise-breaking is wrong independently of a practice of promising,


22.

23. 24.

25.

26.

27.

28. 29.

on his account, only if promises can generate assurances independently of such a practice. I argue that promises cannot generate the requisite assurance in the absence of a practice and, therefore, Scanlon has not succeeded in explaining how promise-breaking is wrong independently of a practice of promising. I am grateful to Mark Tunick for pressing me to clarify this. Kolodny and Wallace also object to Scanlon’s argument that the Principle of Manipulation can explain how promises generate the sort of assurance required to trigger a duty of fidelity (Kolodny & Wallace 2003:139–44). They argue that it cannot provide this explanation because the belief it warrants—a belief in the present intention of the promisor to perform—is insufficient to ‘justify’ reliance by a promisee on that performance. But this objection begs the question, since presumably one of the justifications that a promisee might have for relying on the promised performance is that the promisor has triggered the Principle of Fidelity by convincing the promisee of the promisor’s present intention to perform. The critical question, begged by Kolodny and Wallace, is why this kind of belief cannot trigger the Principle of Fidelity. Joseph Raz invokes a similar example in Raz 1972:99. Scanlon raises a similar example in a footnote. ‘A person who says, “I firmly intend to do X, but I don’t promise to”, gives the kind of warning which makes Principle F inapplicable, and expresses the judgment that, having given this warning, he or she is free to decide not to do X’ (1998:404 n.9). As to why such a warning would render the Principle of Fidelity inapplicable, Scanlon says little. In a previous footnote he suggests that the reason the warning would make the principle inapplicable is that it would preclude the first of that principle’s six criteria from being satisfied (1998:403 n.6). I address this argument below. Intentions are not independent sources of reasons for acting (1998:46; Broome 2001). But one has reason to carry out the intentions one has formed, unless one has reason to reconsider those intentions, as Scanlon himself admits (1998:45–46). ‘In my view, by saying “I promise” one can simultaneously create a new obligation and acknowledge that one is doing so. Indeed, one can do the former in part by doing the latter’ (Scanlon 2001:98 n.16). This answers an objection put by Wallace and Kolodny to an earlier version of my essay. They argue (Kolodny & Wallace 2003:135 n.13) that my view that the value of assurance can support a non-perlocutionary principle of fidelity is mistaken because it depends on the premise that ‘B’s believing that A will do X, as opposed to A’s actually doing X, is an insignificant aspect of the value of assurance’. And this premise is false, since in most cases the value of assurance is ‘rooted in its role in B’s deliberation and planning, a role that it has only in virtue of B’s believing that A will do X’. This argument appears to be based on the authors’ view that the value of B’s believing that A will do X cannot provide valid support to a nonperlocutionary principle of fidelity. I have already shown that this is mistaken. Moreover, as my discussion of ‘expectation-dependent’ reasons shows, I do regard B’s believing that A will do X as a significant aspect of the value of assurance—it is an aspect that bears equally in favour of both the Principle of Fidelity and my alternative non-perlocutionary principle. See text accompanying note 13, above. See above, note 3 and accompanying text.


30. Mark Tunick has argued that the moral principles that govern promise-keeping are non-natural—they are mediated by social practices—and that Scanlon fails to show otherwise. See Tunick 1998:83–90.

REFERENCES Anscombe, G.E.M.. 1978. ‘Rules, rights, and promises’. Midwest Studies in Philosophy, 3, 318–23. Atiyah, P. 1981. Promises, Morals and Law. Oxford: Clarendon Press. Benson, P., ed. 2001. The Theory of Contract Law: New Essays. Cambridge: Cambridge University Press. Broome, J[ohn] 2001. ‘Are intentions reasons? And how should we cope with incommensurable values?’ Morris & Ripstein 2001:98–120. Cohen, M. 1933. ‘The basis of contract’. Harvard Law Review, 46, 553–92. Cupit, G. 1996. Justice as Fittingness. Oxford: Clarendon Press. Den Hartogh, G. 1998. ‘A conventionalist theory of obligation’. Law & Philosophy, 17/4, 351–76. Downie, R.S. 1985. ‘Three accounts of promising’. Philosophical Quarterly, 35/140, 259–71. Finnis, J. 1980. Natural Law and Natural Rights. Oxford: Clarendon Press. Fox, R. & J.DeMarco. 2001. The Immorality of Promising. New York: Humanity Books. Fried, C. 1981. Contract as Promise. Cambridge, MA: Harvard University Press. Hacker, P.M.S. & J.Raz, eds. 1977. Law, Morality and Society: Essays in Honour of HLA Hart. Oxford: Clarendon. Hart, H.L.A. 1994. The Concept of Law. 2nd ed. Oxford: Clarendon Press. Hollis, M. 1998. Trust Within Reason. Cambridge: Cambridge University Press. Honoré, T. 1999. Responsibility and Fault. Oxford: Hart Publishing. Hume, D. 2000. A Treatise of Human Nature, ed. D.Norton & M.Norton. Oxford: Oxford University Press. Kolodny, N. & R.J.Wallace. 2003. ‘Promises and practices revisited’. Philosophy & Public Affairs, 31/2, 119–54. MacCormick, N. 1982a. ‘Voluntary obligations’. MacCormick 1982b:191–211.1982b. Legal Rights and Social Democracy: Essays in Legal and Political Philosophy. Oxford, Clarendon Press. McMahon, C. 1989. ‘Promising and coordination’ American Philosophical Quarterly, 26/ 3, 239–47. Mill, J.S. 1961. On Liberty. New York: Dolphin Books. Morris, C.W. & A.Ripstein, eds. 2001. Practical Rationality & Preference: Essays for David Gauthier. Cambridge: Cambridge University Press. Postow, B.C. 1980. ‘A possible ground of political obligation’. Southern Journal of Philosophy, 18/1, 63–9. Rawls, J. 1971. A Theory of justice. Cambridge, MA: Harvard University Press. Raz, J. 1972. ‘Voluntary obligations and normative powers’. Proceedings of the Aristotelian Society Supplement, 46, 79–102.1977. ‘Promises and obligations’. Hacker & Raz 1977:210–28.1982. ‘Promises in morality and law’. Harvard Law Review, 95, 916–38.1999. Practical Reason and Norms. Oxford: Oxford University Press.


Robins, M. 1976. ‘The primacy of promising’. Mind, 85/339, 321–40. Scanlon, T. 1990. ‘Promises and practices’. Philosophy and Public Affairs, 19/3, 199–226. 1998. What We Owe to Each Other. Oxford: Oxford University Press.2001. ‘Promises and contracts’. Benson 2001:86–117. Searle, J. 1969. Speech Acts: An Essay in the Philosophy of Language. Cambridge: Cambridge University Press.2001. Rationality in Action. Cambridge, MA: MIT Press. Sidgwick, H. 1962. The Methods of Ethics. Chicago: University of Chicago Press. Tunick, M. 1998. Practices and Principles: Approaches to Ethical and Legal Judgment. Princeton: Princeton University Press. Weale, A. 1978. ‘Consent’. Political Studies, 26/1, 65–77.

7 Contractualism and the Virtues JONATHAN WOLFF

One can no longer truly say that virtue theory is the neglected tradition in moral philosophy. I won’t say much about the reasons for its revival, although the reasons for its temporary, though long, decline interest me. Now there are very many things that could be said here. For example, it is often thought that virtue theory requires some sort of teleology, but with the decline of Aristotelian physics and its replacement with the mechanical philosophy of the seventeenth century, notions of function and purpose were given an ever-diminishing role throughout intellectual life (MacIntyre 1982). Alternatively, or in addition, one might see virtue theory condemned through guilt by association with Aristotle. Pursuing this line, it might be predicted that a developing discontent with forms of reductionist naturalism in metaphysics might also give weight to a new respect for Aristotelian themes in ethics. But while I want to paint here with a reasonably broad brush, I want to draw the focus in a little. Thinking specifically now about the relation between virtue theory and the twentieth century, one can see various ways in which virtue theory was out of keeping with the spirit of the age. As we shall see, however, some of the criticisms seem to be somewhat at odds with each other. One general line of criticism, which has several parts to it, starts from a philosophical view about the nature of morality. The second general line is harder to characterise but might be thought to derive from reflection about the model of moral agent that virtue theory offers. There are several related but distinct ways in which virtue ethics has been out of step with modern moral theory. First, a developing consensus—though one now strongly challenged—supposes that there are certain constraints on what is to count as moral behaviour. It presupposes that the central problem of morality is that there is a conflict, to put it crudely, between morality and self-interest. This could be termed as a conflict between altruism and egoism; duty and inclination; or, most helpfully in the present context, concern for others and concern for self. On this view acting morally is a burden, or sacrifice, requiring you to put your own interests to one side. But a moral philosophy of the virtues seems to get this all wrong. It seems to be a morality of self-concern, requiring you to attend to cultivating your own dispositions: a sort of moral grooming and

104 CONTRACTUALISM AND THE VIRTUES

preening. Consequently, it might be thought, the problem with virtue theory is that it is too self-centred, too self-indulgent. Possessing the virtues is meant to benefit their possessor. How can that be a morality? There is more than one issue here, but they have in common the idea of excessive self-concern. We can call this the problem of narcissism. For an example, consider Plato’s account of the just man in Book IV of the Republic. The just man is the person who has achieved a proper balance between reason, spirit and appetite to arrive at a type of inner peace, or harmony. Very nice, we might think, but what has that got to do with being just? On the contrary, a just man, we might argue, is someone who pays his debts and deals honestly with others, whatever state of inner turmoil he is in. Of course it might be true that there is some connection between your inner state of being and how you treat others, but, on this view, treatment of others is primary. So in describing the just man Plato overlooked its essential feature. That is one way that a morality of the virtues fits poorly with twentiethcentury views of the nature and point of morality. A second way involves notions of freedom and responsibility. Here, as Bernard Williams has insisted, modern morality has inherited at least a great deal of respect for Kant’s claim that unless actions are performed for the sake of duty they have no moral worth (Williams 1985:174–96). The less controversial descendent of this view is that unless actions are in some way performed freely, we are not responsible for them and cannot be properly subject to moral assessment. It is thus a necessary condition of the moral evaluation of an agent that the agent is, in the appropriate sense, responsible for his or her actions or states. This will normally require some act of the agent’s will (or lack of act when one should have been expected) at some point in the immediate or pre-history of the action or state. Yet a morality of the virtues seems indifferent to this. Although it requires an individual to develop and consolidate the virtues, should it turn out that some individuals instinctively act as virtue requires, this if anything seems a bonus— an object of further admiration —rather than a deficit. An individual who naturally shows sympathy for others is no less virtuous than someone who has had to learn how to do so. Possibly more so (however, see Hursthouse 1999 and Foot 1978). The case of vice may be somewhat different: cultivating cruelty seems worse than naturally being cruel, but a morality of the virtues typically will not excuse people simply because they were ‘born that way’. This, then, is out of keeping with common assumptions about the role of responsibility in morality, and thus a morality of the virtues will be criticised on the basis that it downplays responsibility. This is the point of conflict with contemporary thought. Note—by way of introduction of the third observation—that I have tended to talk in terms of ‘a morality of the virtues’ rather than ‘virtue theory’. Although sometimes the term ‘theory’ can cover any sort of abstract account, the standard model of a theory is that of scientific theory, where a relatively small number of relatively concisely stated principles can have an immense range of application.

SCANLON AND CONTRACTUALISM 105

In morality utilitarianism is the most obvious version of a theory of this sort, but Kant’s categorical imperative is often, but not always, taken to be another. I think that there is no question that a morality of the virtues leave us short of a fundamental theory of morality, in this sense. I suppose we could say that it does have a fundamental principle: develop the virtues, or, more minimally, act as the virtuous person acts, but this will hardly do. It is no more use than a theory of motion that says: objects often move in characteristic ways. Of course even that is better than silence, but it serves little practical purpose without generous amplification. So here is a third problem: a morality of the virtues lacks fundamental principles. Indeed we might go further. It fails to offer any determinate code of conduct, of a sort that might be read, learnt and followed, like the ten commandments. Perhaps we could go on, but here, then, are three distinct ways in which a morality of the virtues fails to live up to the assumed conditions of adequacy for twentieth-century moral philosophy: it is narcissistic (too self-concerned); it downplays responsibility; and it doesn’t offer a theory. I said that there is also another avenue of criticism, which seems to be in some tension with aspects of the foregoing. The criticisms we have just seen are based on an apparent conflict with a set of philosophical assumptions. The other source, I said, is more connected with the idea of what we can reasonably expect of a moral agent. This is what we could call the problem of the ‘priggishness’ of the virtues. A morality of the virtues is a morality of character, principally, and of action only in so far as it flows from character. To be a virtuous person is to be a person of a certain character: one who possesses the virtues. Somehow the idea of being virtuous has become associated with the idea of possessing certain low church traits, such as chastity, abstinence and sufficient meekness to be a candidate for inheriting the earth. Of course, charity, kindness and (modest) hospitality are there too, but overall, it doesn’t sound much fun. Now, of course, it might be said that acting morally wasn’t meant to be fun, but a morality of the virtues is in a worse position than others in this respect. For other theories require you to act in certain ways, but not necessarily to build traits and dispositions into the fabric of your life. Virtue ethics gives you no time off. Not because, like some forms of utilitarianism, it constantly requires action, but because it requires something deeper than action. It requires you to become a person of a particular type: a rather boring one. At first sight this could hardly be more different from the problems we have looked at above, or, at least, from the first of them: narcissism. How can selfindulgence and priggishness co-exist? Actually there is no problem here. Dostoevsky, in an attempt to undermine the assumptions of the economic rational choice theory of his day, distinguishes self-obsessed behaviour from selfinterested behaviour. Even if it were true that an individual were utterly obsessed with himself, to the exclusion of all others, it would not follow that he would be obsessed with his own interests. He may, in Dostoevsky’s example, be highly self-destructive. Dostoevsky thinks that this is a common part of human


behaviour, however hard it is to explain in standard economic categories. (Dostoevsky 1972:34) Stanley Fish argues that a more moderate form of this is particularly common among college professors, whom, he thinks would generally rather have something to complain about than to have things go well (Fish 1994b). The important lesson is that there is more than one way of being selfcentred. This, then, sets the stage for the current project. I would like to consider whether contractualism can help rescue virtue ethics from the twin vices of narcissism and priggishness (I shall ignore the other criticisms for the purposes of this essay). Now the objection of priggishness may seem particularly weak. There is no reason why virtue ethics should be saddled with the monkish virtues. Other catalogues of the virtues make them far from dull. Indeed when we look at Aristotle (Aristotle 2002), and even more so, Neitzsche, things look very different (Nietzsche 1990). However, this then generates other problems. First, it seems to compound the criticism that virtue morality is far too self concerned. Second, it raises the question of how we are to decide between different accounts of the virtues. We might prefer something somewhere between the monkish virtues and Neitzsche’s account of the great man, but what reasons can we give for one account rather than another? But let us take a step back. The basic form of the problem we have isolated is that virtue ethics is an ethics of self-concern, yet morality primarily requires concern for the interests of others. How is it possible to reconcile the two? Now it seems to me that contemporary contractualism is perfectly suited to address this. Consider Rawls’s account of the two moral powers: the rational and the reasonable. The idea of the rational is that everyone is at least capable of a conception of the good, and has good reason to wish to pursue their particular conception of the good. The idea of the reasonable—or at least one idea of the reasonable—is the preparedness to take the interests of others into account when pursuing one’s own interests. It is the attempt to combine these two that gives contractualism its structure (Rawls 1993:48–54). Now it is interesting that although Rawls concentrates on the virtue of justice, he does say that it may be possible to derive contractualist accounts of the other virtues (Rawls 1971:17). One possible contractualist theory of the virtues runs along the following lines. First, it is accepted that each person is entitled to attempt to live a life that they find appealing. Second, it is recognised that for any individual there are many different types of lives that could be found appealing. Third, it is accepted that morality requires one to moderate one’s pursuit of one’s interests out of regard for the interests of others. Fourth, it is accepted that this will provide a constraint on the types of appealing lives that should be considered ‘accessible’. Fifth, each person, insofar as they are rational and reasonable, will choose the best form of accessible life available to them. Sixth, the life, so chosen, will contain the virtues. In summary, the theory I would like to consider is that the life of virtue is (a component of) the best sort of life one can have, subject to the constraint that


everyone else may live that same type of life. If this is right, then it should be no surprise that virtue ethics includes a strong element of self-concern. It is also no surprise that it should contain constraints. The trouble with the monkish account is that it sets the constraints too high. The trouble with Aristotle and Nietzsche is that they pay too little regard to the constraints. Why do I call this account contractualist? Because it represents the virtues as a possible outcome of agreement: this is how we would agree to develop ourselves if we adopted the right point of view for deliberation. There are, of course, very many complications, but let me mention what I think might be the most immediate one. Just because the virtues are the object of universal agreement (if indeed they are) it does not follow that the ethical code that issues from it must be universal. It could be, for example, that from behind a veil of ignorance rational contractors would agree to different virtues for men and women (Aristotle) or the strong and the weak (Neitzsche). To take a different example, we might decide that it would make for a better society if a small proportion of people were very rich. (It is sometimes argued that this is a spur to innovation.) In reply it could be said that this is acceptable provided everyone has an equal chance of becoming rich. However, we might become convinced that things go even better when there is a dynasty of the rich, who have become accustomed to spending their money the right way. Empirically this is probably false—the ostentatious spending of the nouveau riche is what we need—but it is a possible view. So, by analogy, the accounts of the virtues of Aristotle and Nietzsche could be the object of universal agreement, even if they do not prescribe the same for everyone. So we need to continue to bear this possibility in mind. In consequence I was wrong to say that the theory I am considering is that ‘the life of virtue is (a component of) the best sort of life one can have, subject to the constraint that everyone else may live that same type of life’. Rather, I should have said ‘subject to the constraint that one’s living that life is acceptable to all’, recognising the possibility that we may be prepared to accept some forms of diversity. Let us now attend to the structure and content of the argument framework set out above. It starts with the assumption that everyone is entitled to attempt to live a life that they find appealing: the right to the pursuit of happiness, we might call it. Now this is not self-evident. What, after all, is the source of this entitlement? A religious theory might have an answer, but in a non-religious framework a different approach is needed. Here I will simply take it as a starting point. But it could be denied. In particular it contains an assumption of equality, ruling out elitist assumptions. No group is, at the outset, picked out as more or less entitled than any other group. But what is an ‘appealing’ life? Not necessarily a ‘morally good’ life. While a morally good life might (or might not) be an example of an appealing life, it is not assumed that all forms of appealing life are morally good. A hedonistic life, a life devoted to obsessive pursuit of a single goal, a life devoted to the greatest


possible variety and intensity of experience, or a life following a particular sports team can all, in their way, be or become appealing. The catalogue of possibilities is vast. ‘Appealing’ here means ‘appealing from the point of view of the agent’. However, it might be thought that even on this account the argument is already all but over. Suppose you want to bring your children up so as to have the best life possible. All you care about is that their lives should go as well as possible from their own personal perspective. What should you teach them? Should you teach them to be grasping, selfish, dishonest, cruel, unreliable and so forth? Should you tell them: ‘You’ve got to look after number one in this world, kid; no one else will’? Or should you, as Hare argues, teach them to be kind, generous, trustworthy, respectful and so on? According to Hare, a child brought up this way is much more likely to have a successful life than one brought up with selfish dispositions. This is not to say that dishonesty never pays in a particular instance, but that a settled disposition to act honestly is almost always better for the agent in the long run (Hare 1981:196). Now this is an interesting reflection, and, in part, was the inspiration for the line of argument given here. Yet what it shows immediately is less clear. We need to ask why an individual’s life would go better in this example. Is it because it is intrinsically the best sort of life, or is it, because of the way the rest of the social world is set up, this type of life happens to be rewarded? Consider a domestic analogy. Probably your life will go better if you do a certain amount of washing up. But this is not because there is anything appealing (for most of us) about washing up. If you could find a self-loading, self-emptying dish-washing machine, it is unlikely that you would resist purchasing this merely because your life presently goes better when you do some washing-up. The value of washingup, for most of us, is purely instrumental, normally instrumental to domestic peace. This does not mean that it is always resented, but it is likely to mean it is not sought out. One will not use more pans than strictly needed just in order to have the experience of cleaning them. Now Hare’s argument does not tell us why acting honestly, and so on, pays. One possibility is that it has purely instrumental value, in the sense that it is adopted as a profitable strategy. This makes the life of virtue the solution to a coordination problem: the solution to a multi-person, multi-play, prisoners’ dilemma. At the other extreme, it is possible that it has value entirely independently of the social response to it. This is the case Plato attempts to make out, in accepting the challenge of explaining why it is a better for the agent to be a just man with a reputation for injustice than to be an unjust man with a reputation for justice. Yet there does also seem to be a middle ground. Perhaps acting honestly has conditional value, but its value is not entirely strategic. What I mean is that although it is most valuable under certain conditions—when it is socially rewarded—the value is not merely the value of the social rewards. This is the line we need to develop in order to get the right character to the virtues. For if we are content with the virtues as merely a solution to a collective action problem this gives too thin a sense to the idea that the possession of the virtues benefits


the possessor. The benefit would be entirely contingent on the circumstances in which we find ourselves. On the other hand, if the virtues are considered to be of absolutely unconditional value for the virtuous agent, and of more value than any alternative, then acting non-virtuously would be obviously irrational. Yet this hasn’t always seemed so obvious. So what is the correct way to understand the situation? The idea of conditional, but non-strategic, value seems to require something like the following: (a) Possession and exercise of the virtues will tend to benefit the virtuous agent when social norms are set up to reward and reinforce the virtues. (b) Possession and exercise of the virtues has a value for the agent beyond any social rewards it may generate. (c) Possession and exercise of the virtues has some value for the agent even when it does not generate any social rewards. (d) In at least some cases possessing and exercising the virtues when rewarding norms do not exist will lead to significant costs. (e) It is not always the case that the value of possessing and exercising the virtues will outweigh the disadvantage of possessing those virtues in a world in which they do not receive social reward. Consider now a kind person. Generally being kind is not costless. It requires effort, and may entail material expense. Nevertheless in a decent world it is generally worth being kind. Kind people may be included in collective activities from which the unkind will be excluded. They will gain a reputation which may be of use, although it may, also, make one subject to calls upon one’s kindness. Circumstances vary, but quite likely the instrumental benefits alone of being kind make it worthwhile, provided one lives in the right circumstances. Now, let us imagine asking a kind person why they have cultivated a kind character. Possibly they will deny that they are kind, pointing to examples of people who are even kinder. Possibly they will ask us what we mean. But suppose that they answer that they have become kind because, according to their calculations, kindness is likely to pay off in the longer term. There seems something suspect about this: this level of calculation seems wrong. So should they answer: they have decided to become kind because they have become convinced that one ought to be kind, and so they always try to do the kind thing? Although this is admirable in its way, it also seems unsettling too. Again it is too deliberate, and leaves us wondering whether to accuse them of priggishness, narcissism, or both. So suppose they say: it is good to help people out when you can. This is something we can understand. But what does it mean? Does it mean that they get an inner glow of some sort when they do someone a kindness? Perhaps they do, but can it be that they act kindly because they seek out this inner glow or high, maybe like a fitness fanatic or glue sniffer? But this seems wrong. The kind person does not typically seek ‘kind-highs’. For example, he or she does seek out opportunities where kindness is required (e.g. by taking holidays in poor


countries, although hospital or prison visiting maybe a counter-example). The right thing to say seems to be something like: acting to help other people in the right circumstances is generally satisfying in some sense, although not so satisfying that it is extensively sought out for the sake of such satisfaction. However, it doesn’t really seem a burden, at least in many cases. Furthermore, any possible social rewards will be something like an incidental bonus. So, in any particular case of kind action, a kind person will explain their action by saying ‘I saw that she was in trouble’, or whatever. We will say that they responded to someone in trouble because they are kind. We might also say that they may have become a kind person because it is good to be a kind person. There is a kind of satisfaction to be had in helping people, independently of any benefits the reputation for kindness may bring. Alternatively, now suppose that you live in a world where kind acts are not really understood: if you act to help others then you are thought to be weak, irrational or both. Would you still act kindly? Maybe, maybe not. If you still find it satisfying, then, like someone with an obscure hobby, you might continue even in the face of incomprehension, even derision. (Perhaps this is what has happened with respect to those who still exercise the monkish virtues.) But if the costs are too high, maybe you will change. The key point, though, is that a life of kindness, and I assume of other virtues, is a life with its own satisfactions, independent of social reward. This is enough to make it a candidate for an element of one’s life which can be appealing in its own right. Possessing and acting upon the virtues benefits the possessor of the virtues in this non-strategic sense. So the position I am trying to support sets out by starting from a potential catalogue of appealing lives. For any individual there are many types of dispositions that they may internalise and which may seem very appealing. One might, for example, form the ambition to have a singlefigure handicap at golf, and build one’s whole life around this. It may affect where one lives; the type of work one does; the types of relationships one forms and so on. Although from the outside this may look like a kind of madness, the same can be said about many things. The ambition to write a work of Philosophy that will still be read when one is dead may seem reasonable enough to people reading these words. Making this goal the centre of one’s life in the sense that everything else must be made to fit around it is understandable, but perhaps, in more than one sense, unbalanced. My point, though, is that although we are not infinitely flexible, we can come to understand many possible aspects of life as potentially valuable for their own sake, and that for any of us, there will be many of these which are a real option. A life which includes the virtues is in that catalogue of appealing lives for at least many people. Will there be a unique best, such that it is irrational to settle for less than the best? This partly depends on one’s view of reason and reasoning. But what seems clear is that when one is comparing possibilities from the standpoint of deciding which to follow, one can only have a vague idea of the potential satisfactions and other benefits of a particular style of life. Consequently we are


never in a position to do any more than to make vague estimates of which type of life might be the rational best. For all intents and purposes we treat various options as ‘rationally eligible’ (Raz 2000) and then decide on other grounds. Now I have suggested that adopting the moral point of view is to accept that one must also take the interests of others into account. This may well narrow one’s options dramatically. And this is where contractualism enters the picture. Typically when we think of a contract we have to specify at least four elements: (a) The parties to the contract. (b) The knowledge of the contractors. (c) The motivation of the contractors. (d) The ‘non-agreement’ point, or status quo. Some contractualist theories of morality have turned out rather disappointing. Consider Harman’s position in ‘Moral relativism defended’ (Harman 1975). Here the parties are conceived of as ordinary, real-life agents, who enter the contract with their talents and possessions. The non-agreement point is something like a Hobbesian state of nature. From this we can derive, Harman argues, only the right to non-interference and duties of emergency aid, where the costs of aid are minimal compared to the benefits. For nothing more extensive than this would get the agreement of the rich. To achieve a more substantial moral outcome from the contract we seem to need to assume some moral motivation among the parties. Now this could be mimicked, through ignorance, as in the Rawlsian contract (Rawls 1971), or more directly attributed to the parties, as Scanlon does (1999). I will not here try to delve into such technical details, but the parties are to address the question: insofar as it is under our control, what types of lives would we want people to adopt? At this point we need to revisit the question of whether the outcome of the contract situation yield a universal result, i.e., should it demand the same for all? In this context the question is whether all should accept the same account of the virtues, or whether there is room for variation. Now although my first instinct was to think that we needed a ‘same for everyone’ result, I now think that this is a mistake. What is required is that any differences should be justifiable from a universal perspective. Now there seem to me to be two reasons why we might accept variety in what the theory prescribes for each person. The first, stronger, argument recalls that we began from the proposition that each person is entitled to lead a life that they find appealing. Now, given our varying natures, different types and strengths of virtue will be differently appealing to different people. For some people it may be very hard to live a typically virtuous life, and this will lead to much frustration. Given this, it may well be agreed that less will be expected of these people than those for whom virtue comes easily. Although, of course, we must expect at least


Harman’s minimal morality from all, universal agreement, from, say, behind a veil of ignorance, need not mean a universal standard. Is this unfair? Would those who are required to be most virtuous have a right to complain? No, for two reasons. First, the idea is to equalise burdens in the face of differential natures. Those from whom most is expected are those who will find it easiest to face that challenge. It is an ‘each according to their ability’ type of principle. Second, we should recall that it has been argued that a life of virtue is, for many people at least, an appealing life, or, at least, a component of an appealing life. Being asked to lead such a life should not be a source of resentment. I said that there were two reasons why we might accept diversity. The second, weaker reason, is that it may be a better world overall if we encourage a diversity of character types. But I do not want to develop this argument any further here. I am aware that I have done no more than give an outline of an account. I have not given a list of the virtues, or shown how they would derive from a contract situation, which, in any case, I have said little about. I will not try to make good these defects here. However, I would like to explain how the account so sketched addresses the problem of narcissism and the problem of priggishness. Narcissism is essentially the objection that a morality of the virtues encourages the agent to be too self-concerned. For this reason it is no moral theory. The answer to this is that a life including the virtues is indeed a life that is good for the agent, but it is only one possible good life. It is selected by the agent precisely because, of all the good lives available to that agent, it is the one which does most to take the interests of others into account. So the objection of narcissism is misplaced, for being one-sided. Predictably, the objection of priggishness receives a mirror-image reply. The objection was that virtue theory encouraged people to adopt a narrow, dull life, essentially ignoring one’s own interest in leading a worthwhile life. Again we see this is one sided. Of course the life of virtue does require you to take others’ interests into account. Yet when properly appraised it is, first of all, a life that is appealing in its own right. Or rather, one should be virtuous to the degree that one can be without leading to a life of frustration or misery. That, at least, is the pluralist, permissive account of the virtues that one would expect to derive from the type of contract theory considered here. REFERENCES Aristotle. 2002. Nichomachean Ethics. Oxford: Oxford University Press. Dostoevsky, F. 1972. Notes From the Underground. Harmondsworth: Penguin. Fish, S. 1994a. There’s No Such Thing as Free Speech. New York: Oxford University Press.1994b. ‘The Unbearable Ugliness of Volvos’. Fish 1994a:273–80. Foot, P. 1978. Virtues and Vices. Oxford: Blackwells. Hare, R. 1981. Moral Thinking. Oxford: Oxford University Press. Harman, G. 1975. ‘Moral Relativism Defended’. Philosophical Review, 84/3, 3–22.


Hursthouse, R. 1999. On Virtue Ethics. Oxford: Oxford University Press. MacIntyre, A. 1982 After Virtue. London: Duckworth. Nietzsche, F. 1990. Beyond Good and Evil. Harmondsworth: Penguin. Plato. 1955. The Republic. ed. H.D.P.Lee. Harmondsworth: Penguin. Rawls.J. 1971. A Theory of justice. Oxford: Oxford University Press.1993 Political Liberalism. New York: Columbia University Press. Raz, J. 2000 Engaging Reason. Oxford: Oxford University Press. Scanlon, T.M. 1999. What We Owe To Each Other. Cambridge, MA: Harvard University Press. Williams, B. 1985. Ethics and the Limits of Philosophy. London: Fontana.

Abstracts

Scanlon and Reasons SARAH MARSHALL Scanlon’s account of reasons is essential to his contractualism as a whole, providing an extensive foundation in practical reasoning for his theory. A full understanding of his account of reasons is therefore vital to understanding the nature of Scanlon’s contractualism. With the aim of contributing to such an understanding, in this essay I reconstruct several of Scanlon’s most significant arguments concerning reasons. I focus on two areas: his discussion of the role of desire in practical reasoning and his arguments for the claim that reason judgements should be seen as objective. I conclude that the weakness of one his claims regarding desire may cause substantial problems for his arguments in both of the areas examined. The Magic in the Pronoun ‘My’ SUSAN MENDUS In What We Owe to Each Other, T.M.Scanlon says that any acceptable moral theory must answer what he calls the priority question: the question of why moral value should takes priority over other values, such as the values of love and friendship. In this essay I discuss Scanlon’s answer to the priority question and contrast it with the answer offered by Christine Korsgaard in Sources of Normativity. I argue that each account contains important insights but that neither is completely satisfactory. However, by combining elements of the two accounts we can arrive at a richer understanding of the relationship between morality and other values, and a better explanation of the priority of the moral. Contractualism, Spare Wheel, Aggregation BRAD HOOKER This essay explores the reasons for thinking that Scanlon’s contractualist principle serves merely as a ‘spare wheel’, an element that spins along nicely but

ABSTRACTS 115

bears no real weight, because it presupposes too much of what it should be explaining. The ambitions and scope of Scanlon’s contractualism are discussed, as is Scanlon’s thesis that contractualism will assess candidate moral principles individually rather than as sets. The final third of the paper criticizes Scanlon’s account of fairness and his approach to cases where agents can save either one person or many people. Responsibility and Choice MATT MATRAVERS In this essay I argue that contemporary Anglo-American liberal egalitarianism has at its heart a tension: the goal is to find principles of justice that are fair in respecting the distinction between choice and chance and that do not invoke controversial metaphysical arguments. This is a tension because distinguishing between choice and chance itself requires invoking controversial metaphysical arguments. I proceed by offering, and then examining, the thought that Scanlon’s distinction between ‘attributive’ and ‘substantive’ responsibility offers a route out of the tension described above. The greater part of the essay is taken up with examining Scanlon’s account of responsibility and the distinction between substantive and attributive responsibility. My conclusion is that Scanlon does not offer a compelling account of substantive compatibilism; that his theory does not, therefore, release the liberal egalitarian from the tension; but that this does not show that the direction indicated by Scanlon’s theory is the wrong one. Promises and Perlocutions MICHAEL PRATT This essay is a critical analysis of T.M.Scanlon’s contractualist account of promising and promissory obligation. After situating Scanlon’s account within one of two broad schools of thought on the topic—the ‘perlocutionary’ school— I argue that his account fails to overcome a fatal circularity that plagues all perlocutionary theories of promise. I go on to argue that Scanlon’s contractualist moral theory will support an alternative, non-perlocutionary theory of promise that is not susceptible to this logical difficulty. Contractualism and the Virtues JONATHAN WOLFF An ethics of the virtues has been accused of being too directed to the interests of the moral agent to constitute a moral theory; yet at the same time the notion of a virtuous agent conjours up an image of a joyless prig. This essay attempts to explain and defuse these criticisms. It analyses an ethics of the virtues as being motivated by an attempt to combine the good life for the agent with a proper concern for others, in that a virtuous life can be regarded as the most reasonable


good life one can have. The device of contractualism is used to develop more details of this picture.

Index

aggregation, 69–75 Anderson, Elizabeth, 77, 91 Anscombe, Elizabeth, 116, 118 Aristotle, 16, 120, 123–5, 132 Ashford, Elizabeth, 70, 75 Atiyah, P, 99, 116, 118 assurance, 98–103, 111–14

Hacker, P.M.S., 118 Hare, Richard, 126, 132 Hare, Robert, 91 Harman, Gilbert, 130, 132 Hart, H.L.A., 116, 118 Harris, Robert, 78–80, 83, 86–7, 89 Hirose, I., 72, 76 Hobbes, Thomas, 130 Hollis, Martin, 94, 118 Honoré, Tony, 116, 118 Hooker, Brad, 7–8, 61, 75, 76 Hume, David, 2, 93, 94–5, 99–100, 116, 119 Hurley, Susan, 31, 32 Hursthouse, R., 122, 132

Baron, Marcia, 37–8, 46, 52 Barry, Brian, 36–7, 52, 75, 77, 91 Becker, Lawrence, 36, 52 Benson, P., 118 Blackburn, Simon, 75 Broome, J., 118 Cohen, G.A., 9, 12, 78, 80, 91 Cohen, M., 103, 118 Coleman, J., 91 Cullity, Gareth, 75, 76 Cupit, Geoffrey, 94, 118

Ibsen, Henrik, 6, 42–3, 52 Kamm, Francis, 70, 73, 74, 75, 76 Kant, Immanuel, 2, 43, 121, 122 Kelly, Paul, 52 Kolodny, Niko, 101–103, 111–112, 116, 117, 118, 119 Korsgaard, Christine, 5–7, 12, 35, 38–9, 40, 43–5, 46, 49, 50, 52 Kumar, R., 72, 76 Kutz, Christopher 79–80, 91 Kymlicka, Will, 57–8, 76

DeMarco, J., 99, 116, 118 Den Hartogh, G., 110, 118 Dostoevsky, Fyodor, 123, 132 Downie, R., 94, 118 Dworkin, Ronald, 76, 78, 87, 89, 91 Finnis, John, 103, 116, 118 Fish, Stanley, 123, 132 Fried, Charles, 116, 118 Foot, Phillipa, 122, 132 Fox, R., 99, 116, 118

Lamb, Charles, 36 Locke, John, 36, 52 MacCormick, Neil, 99, 116, 119 MacIntyre, Alasdair, 120, 132 MacLeod, C., 4, 12

Gauthier, David, 48, 49, 52 Godwin, William, 5, 34–7, 50, 52 117


Marshall, Sarah, 2–3. Matravers, Matt, 9–10, 12, 91 McGinn, Colin, 75, 76 McMahon, C., 116, 119 McMurrin, S., 92 Mendus, Susan, 5–7, 11 Mill, John Stuart, 4, 116, 119 Morris, C.W., 119 Nagel, Thomas, 21, 32, 54, 76 Nietzsche, Friedrich, 123–5, 132 Norcross, A., 74, 76 Otsuka, Michael; 72, 76 Parfit, Derek, 57, 61, 74, 75, 76 Pettit, Philip, 75, 76 Plato, 121, 127, 132 Postow, 116, 119 Pratt, Michael, 9–10. prisoner’s dilemma, 126 promising, 93–115; perlocutionary theories of, 94, 95, 105; volitional theories of, 93, 110, 114 psychopathy, 84–5, 89 Rachels, James, 32 Rawls, John, 1, 11, 48–9, 52, 56–58, 68, 75, 76, 78, 80, 92, 116, 119, 124, 130, 132 Raz, Joseph, 75, 76, 116, 117, 119, 129, 132 reason, as basic in motivation, 13–14; objective, 25–31 reasonableness, 16, 124 Reibetanz, S., 72, 76 responsibility, 9–10, 77–90, 122; and self-governance, 82–3, 85; attributive, 81–2, 83, 84; attributive and substantive compared, 81, 83, 87, 90; substantive, 85–8, 90 Ridge, Michael, 75, 76 Ripstein, Arthur, 119 Robins, Michael 110, 119 Rousseau, Jean-Jacques,

Scanlon, Thomas M.; 12, 31, 32, 52, 75, 76, 91, 92, 116, 117, 118, 119, 132; account of excuses in, 82; analogy between belief and intention formation, 3, 17–20; and compatibilism, 9, 82–3, 88, 90; and desires, 17, 20–25; and irrationality, 15; and metaphysical minimalism, 1–2, 81, 89; and motivation, 1, 13, 20–25, 30–31, 55, 57; and reasons, 2–3, 13–31, 39; and responsibility, 9–10, 77–90; and scope of contractualist morality, 1, 7, 53–6; and wrongness, 4; buck passing account of value, 3–4, 97– 8; conditions of agreement in, 56; considers moral principles in ‘sets’, 62– 6; hypothetical nature of contractualism, 8, 56–7; on aggregation, 69–75; on fairness, 8, 66–8; on friendship, 5–6, 18, 34–6, 39–51, 98; on promising, 10, 93–115; judgement sensitive attitudes, 14, 27–9; Principle of Fidelity, 95–8, 101–109, 110–11; Principle of Manipulation, 106–109; priority of the moral, 5–7, 33–51; reasonableness in, 16; ‘spare wheel’ (circularity) objection to, 8, 57–60, 62–3, 66, 69, 74; value of choice, 85–6 Schoeman, Ferdinand, 92 Searle, John, 116, 119 Sen, Amartya, 12, 52, 76, 92 Shapiro, Scott, 91 Sidgwick, Henry, 116, 119 Singer, Peter, 76 Stephen, James Fitzjames, 88–9, 90, 92 Stocker, Michael, 51, 52 Stratton—Lake, P., 75, 76

119

Thomson, Judith Jarvis, 59–61, 75, 76 Tunick, Mark 117, 118, 119 utilitarianism, 67–9, 122–3 Virtue ethics, 10–11, 120–31; and contractualism, 123–5, 129–31 Wallace, R.Jay, 75, 76, 101–103, 111–12, 116, 117, 118, 119 Watson, Gary, 78, 83, 91, 92 Weale, Albert, 116, 119 Williams, Bernard, 12, 33–4, 52, 92, 121, 132 Wolff, Jonathan, 10–11 Wrongness, 1, 60–62

Scanlon and Contractualism

Mitchel Scanlon

Mitchel Scanlon

Essentials of Anatomy and Physiology Scanlon

Contractarianism Contractualism (Blackwell Readings in Philosophy)

Pride And Prejudice And Zombies

Pride and Prejudice and Zombies

Pride And Prejudice And Zombies

Lions And Tigers And Snares

Pride and Prejudice and Zombies

Pride and Prejudice and Zombies

Pride and Prejudice and Zombie

Pride And Prejudice And Zombies

Pride and Prejudice and Zombies

Theogony and Works and Days

Pride And Prejudice And Zombies

PRIDE AND PREJUDICE AND ZOMBIES

Humans and Demons and Elves

Pride and Prejudice and Zombies

Pride and Prejudice and Zombies

Pride and Prejudice and Zombies

Pride And Prejudice And Zombies

Consciousness and Persons: Unity and Identity (Representation and Mind)

Milk Consumption and Health (Food and Beverage Consumption and Health)

Abuse, Torture and Trauma, and Their Consequences and Effects

Man and the Beasts (Medieval and Renaissance Texts and Studies)

Environmental and Health Risk Assessment and Management: Principles and Practices

The Media and Hurricanes Katrina and Rita: Lost and Found

Normal and Abnormal Fear and Anxiety in Children and Adolescents

Scanlon and Contractualism

Mitchel Scanlon

Mitchel Scanlon

Essentials of Anatomy and Physiology Scanlon

Contractarianism Contractualism (Blackwell Readings in Philosophy)

Pride And Prejudice And Zombies

Pride and Prejudice and Zombies

Pride And Prejudice And Zombies

Lions And Tigers And Snares

Pride and Prejudice and Zombies

Pride and Prejudice and Zombies

Pride and Prejudice and Zombie

Pride And Prejudice And Zombies

Pride and Prejudice and Zombies

Theogony and Works and Days

Pride And Prejudice And Zombies

PRIDE AND PREJUDICE AND ZOMBIES

Humans and Demons and Elves

Pride and Prejudice and Zombies

Pride and Prejudice and Zombies

Pride and Prejudice and Zombies

Pride And Prejudice And Zombies

Consciousness and Persons: Unity and Identity (Representation and Mind)

Milk Consumption and Health (Food and Beverage Consumption and Health)

Abuse, Torture and Trauma, and Their Consequences and Effects

Man and the Beasts (Medieval and Renaissance Texts and Studies)

Environmental and Health Risk Assessment and Management: Principles and Practices

The Media and Hurricanes Katrina and Rita: Lost and Found

Normal and Abnormal Fear and Anxiety in Children and Adolescents

Recommend Documents